Proceedings of

the Sixth International Workshop

on Coastal Disaster Prevention

December 1 - 2, 2009
Holiday Inn Bangkok, Bangkok, Thailand

Port and Airport Research Institute, Japan

Development Institute of Technology, Japan
Ministry of Land, Infrastructure, Transport and Tourism, Japan
Department of Civil Engineering, Chulalongkorn University, Thailand
Climate and Disaster Science and Technology Center, Ministry of Science and Technology, Thailand

CONTENTS
Tsunami Research - Its past, present and near future - ..................................................1
Nobuo Shuto

Tsunami Risk Assessment and the Planning and Implementation of Strategic Mitigation
Measures- Case Study City of Galle..............................................................................25
Samantha Hettiarachchi, Saman Samarawickrama and Nimal Wijeratne

Real Time Monitoring for Mega Thrust Earthquakes and Tsunamis around the Nankai
Trough Southwestern Japan - Towards to understanding mega thrust earthquakes and
disaster mitigation - .......................................................................................................37
Yoshiyuki Kaneda, Katsuyoshi Kawaguchi, Eiichiro Araki, Hiroyuki Matsumoto,
Takeshi Nakamura, Shinichiro Kamiya, Keisuke Ariyoshi, Takane Hori, Hide Sakaguchi,
Maddegedara Lalith and Toshitaka Baba

Tsunami Warning System in Japan................................................................................47

Kenji Nakata

Experimental Study on Tsunami Power ........................................................................51

Taro Arikawa

Necessity of Advanced Tsunami Damage Index ...........................................................61

Koji Fujima

Detailed and Real-time Estimation Methods of Tsunami..............................................71

Takashi Tomita, Daisuke Tatsumi and Kazuhiko Honda

Tsunami Evacuation Simulation in an Urban Area of Japan .........................................89

Kentaro Kumagai

Tsunami Disaster Reduction in Japanese Ports and Harbors.........................................93

Naokazu Takata

Recent Progress on Tsunami Disaster Mitigation in Indonesia ...................................107

Subandono Diposaptono and Enggar Sadtopo

Proceedings of the 6th International

Workshop on Coastal Disaster Prevention,
Bangkok, Thailand, December 1-2, 2009

TSUNAMI RESEARCH
ITS PAST, PRESENT AND NEAR FUTURE

Nobuo SHUTO
ARISH, Nihon University, Tokyo, Japan, shuto@nihon-u.ac.jp

Abstract
Each time when a tsunami occurred, research seeds were found and tsunami research made
progress.

After the 1933 tsunami, a small scale experiment began to understand the

generation mechanism of tsunamis.

theoretical works.

Before 1960, the mainstream of tsunami research was

Generation of tsunamis and T-waves was predicted.

Need of the

dispersion term was theoretically recommended in the analysis of far-field tsunamis.

In

1960, large-scale experiments and numerical simulation started. After 1970s, numerical
simulations with the shallow-water theory have been growing with remarkable speed. The
Mansinha-Smylie method provided a way to determine the initial profile of homogeneous
fault model, which is being replaced by heterogeneous fault model.

In 1983, a video

clearly showed soliton fission in the sea that requires the non-linear dispersive equation.
In 1993, a tsunami trace in a narrow valley urged the introduction of the 2D/3D hybrid
simulation. In addition to the analysis and simulation of tsunamis, there are many needs
from a viewpoint of disaster mitigation.
Keywords: Numerical simulation, initial profile, propagation, run-ups, damage assessment

1. Introduction
From the birth to the coastal effects of tsunamis, each research topics are described in order of time.
In Section 2, causes of tsunami are summarized. Although sub-marine earthquakes are the major
subject in the present paper, another cause such as volcanic activity should be paid attention, if we
consider the Krakatau eruption that claimed over 36,000 lives.
In Section 3, generation of tsunamis and related problems is discussed. A sub-marine earthquake
generates not only tsunami but also T-wave.

The tsunami itself generates infrasound.

Static and

dynamic generation mechanisms are explained.

In Section 4, the method to estimate tsunami initial profile is described.

Before 1971, the inverse

propagation diagram was used. With the Mansinha-Smylie method, the fault parameters determined
from seismic data are used to estimate tsunami initial profile. After several experiences, assumption of

-1-

homogeneous fault model is being replaced by heterogeneous fault model.

In Section 5, tsunamis in the deep sea are discussed. For a near-field tsunami, the liner long wave
theory is applicable while for a far-field tsunami the linear Boussinesq equation is necessary if the Kajiura
number pa is smaller than 4.
Section 6 is for tsunamis in the shallow sea.

The fundamental equations are the shallow-water

theory with bottom friction included. If a tsunami front shows cnoidal bore, dispersion effect should be
taken into consideration. There is no theory applicable to edge bores, running along the shoreline with
breaking front.
Section 7 is for tsunamis on land.

Run-up front needs q special assumption of the moving boundary.

For a special topography, three dimensional equations are required.

To deal with this condition, a 2D/3D

hybrid simulation is introduced. For a judgment whether or not a numerical simulation is carried out
satisfactorily for practical application, Aida measures K and are used.
In Section 8, four topics important from a practical point of view are selected; damage to houses and
buildings, impact of waves and floating materials, erosion due to tsunami-induced currents, and tsunami
control forests.
Section 9 is for research topics to be solved in the near future. Section 10 is concluding remarks.

2. Causes of Tsunamis
2.1 The oldest document
Thucydides, a Greek historian, was the first person who recorded a tsunami and thought that the
tsunami was generated by an earthquake. In summer of 426 B.C. during the Peloponnesian war, the sea
receded after an earthquake, and then a huge wave hit the city of Orobiae at the northwestern coast of
Euboea Island. A part of the city subsided and became the sea. At the island of Atlanta at the other side
of the strait was also hit by the tsunami.

A part of Athenian fortifications was swept away.

Thucydides

considered that the full force of the earthquake drew the seawater from the shore and then the sea
suddenly swept back again even more violently.

2.2 Submarine Earthquake

Most of the causes of tsunamis are submarine earthquakes. Not the ground shaking but the vertical
displacement of sea-bottom generates a tsunami.

The greater an earthquake is, the larger is the vertical

displacement; hence the larger tsunami is generated.

exception is a tsunami earthquake.

This is a rule for a tsunamigenic earthquake. An

Much larger tsunamis than expected from earthquake seismic waves

can be generated.

2.3 Landslide
Landslides can also generate tsunamis. Lituya Bay in Alaska repeatedly experienced huge local
tsunamis in 1958, 1936, 1899, 1853-1854, and probably in 1900. This bay is about 11 km long, 1 km
wide and 169 m deep. On July 10, 1958, an earthquake caused 30 million cubic meters of rocks
weighing 90 million tons to slide from the northern shore from an average height of 600 m with the
dimensions of 700 m to 900 m and the average thickness of 90 m.

-2-

The slide forced the water surge up to

the height of 520 m on the opposite shore. Then, the water ran down into the bay to form a huge tsunami
higher than 30 m in the bay (Lander, 1996).

2.4 Volcanic Activity

Another mechanism is volcanic actions. In May 1883, the volcanic eruption of Krakatau in the
Sunda Strait between Java and Sumatra Islands began. On August 27, a giant tsunami was generated.
Because of thickly falling ashes, no one could see the tsunami offshore. When coastal residents noticed
the white cap of the tsunami, the tsunami was just in front of them.

There was no time for evacuation.

It claimed more than 36,000 human lives. Its maximum runup was higher than 30 m Simkin and
Fiske).

3. Generation of Tsunamis, T-waves and Infrasound

3.1 Tsunami Earthquake
The Meiji Sanriku Great Tsunami on June 15, 1896, opened the gate to the tsunami science. It was a
typical tsunami earthquake. Many coastal residents felt no ground shaking and did not try to evacuate.
A giant tsunami with the maximum runup height of 38 m hit and claimed 22,000 lives. Since the ground
shaking was too weak, scholars considered that the cause might be a submarine landslide or an action of a
submarine volcano. Imamura (1899) checked the wave length on the tide records that reflected the size
of the tsunami source area, and found that the source area should be wider than several ten kilometers, too
wide for the possible size of a landslide or an eruption of volcano. He concluded the submarine
earthquake was the cause.
To understand the mechanism of a tsunami earthquake, we have to wait until 1977 when Tanioka,
Ruff and Satake (1997) proposed an assumption of the horst-graven topography.

3.2 Tsunamigenic Earthquake

(1) Static Generation
On March 3, 1933, the Showa Great Sanriku Tsunami hit after a strong ground shaking. The modern
tsunami science and technology began with this tsunami.

Takahashi (1934) began hydraulic

experiments of tsunami generation efficiency with apparatus

shown in Fig.1.

Speed and stroke of the piston were adjusted.

He found that a large stroke with the high speed of the piston
movement yielded high waves.
Kajiura (1970) discussed theoretically the generation
efficiency based upon the long wave approximation.

The

energy transfer from the sea bottom to the water was examined
in relation to the duration of the bottom movement. If the
duration is less than several minutes, the deformation may be
considered to be abrupt as far as the tsunami is concerned.
This idea leads to the static theory of tsunami generation; that

Figure 1: Takahashis experimental

is, the final vertical displacement of sea bottom surface is

setup

-3-

equal to the tsunami initial

profile.

(2) Dynamic Generation

Omachi et al. (1999) carried out a dynamic analysis of tsunami generation and showed that the
tsunami height could be larger than the static displacement.

If the period of the Rayleigh wave

approaches the water wave period that is determined by the water depth, the tsunami is amplified.

For a

tsunami generated in the shallow sea, this mechanism may become non-negligible.

3.3 T-Wave Generation

Kajiura (1970) also predicted that if the movement is completed in a few second, the energy transferred
to the compressional water waves (T-wave) might be larger than the tsunami energy.

Iwasaki (1992)

discussed the relation between the movement of sea bottom surface and characteristics of T-wave. The
generation of the compressional water waves was proved by a record of the 2004 Indian Ocean Tsunami
(Okal et al., 2007).

3.4 Infrasound Generation

When a tsunami appears on the sea surface, this vertical displacement compresses the air and
generates the compressional air waves (Iwasaki, 1992; Izumiya and Nagaoka, 1994). This infrasound, not
perceived by human, propagates with the speed of 340 m/s faster than tsunamis.

When the 2004 Indian

Ocean Tsunami hit Sri Lanka and Thailand, elephants detected this low-frequency sound and evacuated
from the sea shore. This sound also recorded on a CTBT microphone (Garces, M. et al., 2005).

4. Tsunami Initial Profile

4.1 Inverse Propagation Method
Before 1971, the inverse propagation method was used to estimate the initial profile. Starting from a
tide gauge station, wave fronts (inverse wave fronts) are drawn, on assuming the propagation velocity of
linear long waves. The inverse wave front that corresponds to the time between the earthquake and the
arrival of the tsunami, gives the position from any place of which the tsunami can starts and arrives at the
tide station. An envelope of the inverse wave fronts from available tide gauge stations is the landward
boundary of the tsunami source area. From the profile of the first wave, wave length is calculated to
determine the width of the tsunami source, i.e., seaward boundary of the tsunami source. Horizontal
distribution of rise and fall of the water surface is also estimated, using tide records.

4.2 Use of Fault Parameters

Mansinha & Smylie (1971) introduced a way to calculate the sea-bottom displacement based on fault
parameters. From analysis of the seismic wave record, a fault movement is described by geometrical
characteristics (location, depth, strike-, dip- and slip-angles of the fault plane), physical characteristics
(length, width and dislocation of the fault plane) and dynamic characteristics (rupture direction, rupture
velocity and rise time of the fault movement).

With fault parameters (except for dynamic

characteristics), the static displacement of sea bottom surface can be computed, assuming that a fault
movement occurred in a semi-infinite elastic, homogeneous body.

-4-

This assumes a homogeneous

movement in a fault plane that leads us to a simple tsunami profile: one crest and one trough in the area of
the generation.

More complete study was carried out by Okada (1985) with correction of misprints in

Mansinha & Smylie paper.

If the fault movement is not large, this estimation works well. An example is shown in Figs. 2 and 3.

Figure 2: The source of the 2004 Kii Hanto-oki

Figure 3: Comparison between the measured

tsunami.

and measured.

The main shock of MJ=7.4 occurred at 23:57 off the Kii Peninsula generated a tsunami (Fig.2).

The

record of a GPS wave gauge, after 150 second moving average, shows the tsunami as in Fig.3. The upper
line in Fig. 3 is the computed by Koike et al. (2005).

4.3 Actual Examples

However, for a big fault motion, the Mansinha-Smylie method does not always give satisfactory
results. Fig. 4 shows the initial profiles of the 1993 Hokkaido Nansei-Oki tsunami (MW=7.8) proposed
by different researchers. The left was computed based upon the Harvard CMT solution assuming one
fault plane. The middle was by Kikuchi who used the seismic records near the epicenter and assumed
two fault planes. Both used only seismic data. If started with these initial profiles, the tsunami was not
explained.
The DCRC of Tohoku University repeated trials 24 times to find DCRC-17a that satisfies earthquake,
tsunami and topographic change (Takahashi et al. 1995).

Figure 4: Initial profile proposed by Harvard (left),

Figure 5: Measured profile of the

Kikuchi (middle) and DCRC (right).

1964 tsunami.

-5-

Only one example of the measured initial tsunami profile is found in the case of the 1964 Great
Alaska earthquake.

Figure 5 is obtained by Plafker (1965). A section along the line A-A is shown in

the lower figure. There is a gentle wavy deformation 450 km long with one crest and one trough. Its
wave height is about 6 m.

Near the crest, there is a sharp rise about 6 m high and about 30 km wide at

its base. We can compute this long deformation with the Mansinha-Smylie method but not the sharp rise,
a result of a sub-fault developed in the accretionary prism, although this rise makes the tsunami height
almost double.

4.4 Heterogeneous Fault Model

There have been several efforts to obtain more realistic initial
profile using seismic records (for example, Fukuyama and Irikura,
1986) or tide records (Satake, 1989).

Inversion introduced by

Satake (1989) is as follows. The fault plane is first divided into

several subfaults and the deformation on the ocean bottom is
computed for each subfaults with a unit amount of slip. Using
this as an initial condition, tsunami waveforms are numerically
computed on actual bathymetry.

The observed tsunami

waveforms are expressed as a superposition of the computed

waveforms as follows,
Aij(t)xj = bi(t)

(1)

Where Aij is the computed waveform, or Greens function, at the

ith station from the jth subfault; xj is the amount of slip on the jth
subfault; and bi is the observed tsunami waveform at the ith station.

Figure 6: Satakes heterogeneous

fault model.

The slip xj on each subfault can be estimated by a least-squares inversion of the above set of equation.
Figure 6 is for the 1968 Tokachi-oki tsunami by Satake.
Recent development of asperity accelerates the need of the heterogeneous fault estimation.

5. Propagation in the Deep Ocean

5.1 Assumption of Long Waves
When generated, a tsunami has a wavelength several tens kilometer long that is much longer than the
water depth.

For example, the average depth of the Pacific Ocean is about 4.2 km.

This water wave is

categorized as long waves, for which the hydrostatic water pressure is a good, first-order approximation.
The initial height of the tsunami is several meters, quite small compared to the water depth and the wave
length. This water wave belongs to the small amplitude waves and is analyzed with the linear long wave
theory.
In the sea shallower than 200 m, a tsunami height becomes of finite magnitude, not infinitesimally
small compared with the water depth. The shallow-water theory, the first-order non-linear long wave
equation, is used. If its water surface develops to have a non-negligibly wavy shape, the effect of
surface curvature that is called as the dispersion effect should be taken into consideration.

-6-

5.2 Phase Velocity Under the Long Wave Approximation

Corresponding to this process of development, the phase velocity is expressed as follows.
(2)
The first term is for a linear long wave, determined only by the gravitational acceleration g and the water
depth h. The second term expresses the contribution of the finite amplitude . The higher point
propagates faster, and the wave front becomes steeper approaching the shore.
effect of the local curvature of the water surface.

The third term is the

A wave crest for which the curvature takes a negative

value propagates slower. Once the second and third terms cancel out each other, a long wave of finite
amplitude can propagate keeping its profile. Typical example is a solitary wave.

5.3 Dispersion Effect for Far-field Tsunamis

For a tsunami in the ocean deeper than 200 m, the effect of the finite amplitude is always negligible
but the dispersion effect is, sometimes, not negligible.
component propagates with lower phase velocity.

The dispersion effect means that high frequency

This difference, although very small, results in

non-negligible deformation in wave profile, if the travel distance is long as in case of a far-field tsunami.
A parameter pa proposed by Kajiura (1970) is used to judge whether the dispersion effect should be
included or not:
pa = (6h/R)1/3(a/h)

(3)

where h is the water depth, a the horizontal dimension of the tsunami source and R the distance from the
source. If pa is smaller than 4, the dispersion effect should not be neglected and the linear Boussinesq
equation that includes the first-order effect of the phased dispersion should be used. The Coriolis effects
are also required. In placed of the Cartesian coordinates, the latitude-longitude coordinate system is
used.

5.4 Use of the Imamura Term

The difference equation of the linear Boussinesq equation becomes complicated because of the
dispersion term and needs more CPU time for numerical simulation. Imamura et al. (1990) introduced a
clever way to reduce the computation time but keeping the same accuracy. The linear Boussinesq
equation is as follows and the third term is the dispersion term.
(4)
When the linear long wave equation is expressed by a difference equation as follows, the first term of
truncation error takes the similar form to the physical dispersion term.
(5)
where K is the Courant number, the ratio of the physical wave propagation velocity C0 to the artificial one

-7-

(x/t). Select the spatial grids and time steps so as to make the coefficients of the third terms in both
equations the same, and then the linear long wave difference equation gives the same result as the
physically correct linear Boussinesq equation.

5.5 Dispersion due to the Coriolis Force

The Coriolis force has a dispersion effect (Imamura et al, 1990). The phase velocity of the linear
long waves is modified to
C=(gh)1/2[1-(f/)2]1/2

(6)

where is the wave number and f is the Coriolis coefficient. For a huge tsunami such as the 1960
Chilean tsunami, the Pacific Ocean behaves like a small pond. Average water depth of the Pacific
Ocean 4.2 km, gives the tsunami a propagation velocity faster than 730 km/hr. The tsunami traveled
17,000 km from the source off Chilean coast to Japan within 23 hours. It started toward Japan with the
crest at its front but arrived at Japan with a big ebb. The first crest became unrecognizably small and the
following trough began to grow near the Hawaiian Islands. This change was the result of the dispersion
due to the Coriolis Effect.

5.6 Topography Effects

(1) Scattering by Sea Mounts
Traveling over the ocean, a far-field tsunami is affected much by the bottom topography. When a
tsunami passes a sea mount, scattered waves are induced according to the scale of the sea mount, and
propagate as the concentric circles from the mount.
sea mount or rise.

Tsuji (1977) discussed the wave scattering over a

In case of the 1960 Chilean tsunami, 40% of the incident tsunami was lost due to

wave scattering during its travel from the coast of Chile to Japan. Mofjeld et al. (2001) analyzed the
scattering capability of the topography in the Pacific Ocean, and found a narrow band of strong scattering
running across the ocean from the northwest (Emperor Seamount Chain) to the southeast (Easter Island
Fracture Zone).

Based upon their results, they recommend that numerical models of trans-Pacific

tsunamis must resolve the effects of the small-scale topography in order to accurately simulate their wave
patterns and amplitudes.
(2) Transmission by Submarine Ridge
A submarine ridge acts as a good wave guide. The shallower the water depth is, the slower is the
wave propagation. Waves change their propagation direction toward the shallower ridge crest.

This

refraction causes concentration and effective transportation of tsunami energy along an ocean ridge. The
energy of the 1996 Irian Jaya tsunami was transported to Japan although Japan is not located on the major
direction of initial tsunami energy radiation. A numerical simulation clarified that the South-Honshu
ridge acted as an effective wave guide (Koshimura et al., 1999).
(3) Edge Waves on the Continental Shelf
Tsunamis behave as edge waves on the continental shelf. When a far-field tsunami is incident
almost parallel to the shoreline, multiple reflected waves are excited and are trapped on the slope. The

-8-

interference between these reflected waves and incoming incident waves could be a cause of unexpected
tsunami amplification on the shelf (Koshimura, 2002).

In case of a near-field tsunami, edge waves

through different propagation path determine the tsunami.

Radiated from the source of two-dimensional

extension, the tsunami front enters the slope with locally different incident angle. The first wave at a
site is the component that propagates relatively deep area, and then other components of high energy
arrive later after a slow propagation on the slope along the shore.

6. Tsunamis in the Shallow Sea

6.1 Amplification Mechanism
In the shallow sea where a tsunami becomes of non-negligible wave height compared to the water
depth, the effect of finite amplitude is taken into consideration. The tsunami height H varies according
to the Green Law that assumes the constant energy transmission rate as the water depth h and the width of
a channel B change. The former is the shoaling effect and the latter is the focusing effect.
(gh)1/2H2B=constant,

H ~ h-1/4B-1/2

(7)

Another important amplification phenomenon that occurs in relation to topography is resonance in

bays and harbors. If a bay satisfies the following relation with an incident tsunami, the oscillation at the
head of bay becomes quite high.
4l = (gh)1/2T

(8)

where l is the length of the bay and T is the wave period of the tsunami.

6.2 Tsunami Front

The shallow-water equations applicable to this long wave include the first-order non-linear,
convection terms.

The water pressure is still assumed to be hydrostatic. The non-linear effect in the

phase velocity makes the higher rear part to go forward faster, then tsunami profile becomes step-wise,
and the tsunami front steepens. This is a bore.
When the steep bore front continues breaking, it is the breaking bore and is solved with the
shallow-water theory.
When the effect of the local water surface curvature prevents the breaking, a train of short-period
waves appears and develops at the front. This is the soliton fission that was clearly recorded on videos
(NHK, 1983).

Tsunamis in river often take this type of bore. This bore is called as a wavy bore or a

cnoidal bore, according to cn(x, k) of the Jacobian elliptic function that is a periodic solution of the
non-linear dispersive long wave equation.

As the modulus k of the cn-function tends to 0, a cnoidal

wave becomes a solitary wave. The water pressure is modified by the centrifugal acceleration from the
hydrostatic distribution. For a tsunami initial profile that has high frequency components, the dispersion
terms should be taken into consideration, even if the source area is close to land.

6.3 Edge Bores

A breaking bore propagating along the shore is an edge bore. This was also recorded on video
published by NHK (1983). There is no theory applicable to the edge bore. Uda et al. (1988) carried
out large scale hydraulic experiments and found several interesting phenomena.

-9-

An edge bore

propagates sometimes following the ordinary refraction law and sometimes neglecting the topography.
A small difference in the side boundary condition introduces a big difference in wave profile, thus
suggesting also a big difference in wave force.

6.4 Non-linear Dispersive Equations

There are many kinds of non-linear dispersive equations.

Different expressions are resulted from the

different basic parameters for perturbation. The Peregrine (1967) used the section-averaged velocity as
the fundamental variable. Madsen & Sorensen (1992) selected the depth-integrated variable, i.e. the
discharge flux.

Equations by Iwase et al. (1998) are composed of the same terms as those of the

Peregrine equations but expressed in terms of the discharge flux. Other perturbations are possible, using
the bottom velocity (Mei & LeMehaute, 1966) or the surface velocity (Dingemains, 1973). Iwase et al.
(2002) carried out numerical computation using different 8 equations and compared the results with the
theoretical prediction and hydraulic experiments. The best agreement was obtained with the Iwase et al.
equations and the Madsen & Sorensen equation. The former, in other word, the Peregrine equation in
terms of the discharge flux is more convenient because it is simpler than the latter.
(9)
The use of the Iwase et als modified Peregrine equation is recommended for numerical simulation of a
near-field tsunami from its source in deep sea to the shallow water region.

7. Tsunami Run-up on Land

7.1 Theoretical Approach
Although our most concern is tsunamis in the near-shore zone and on land where human activity is
intense, it is not easy to simulate the tsunamis and their effects.

Before the numerical computation with

electronic computer was introduced, such simple but basic problem as one-dimensional run-up on a
sloping beach needed a skilful method of the Carrier-Greenspan transform (1958) for waves of finite
amplitude.

The same problem was solved with the linear long wave theory expressed in the Lagrangian

description (Shuto, 1967). These kinds of theoretical consideration can not be applied to solve any
practical problems but only supply data necessary for examination of the numerical scheme.

7.2 Numerical Simulation with the Shallow-Water Theory

The shallow-water equations are successfully applied in the numerical simulation of tsunami run-up
even if tsunamis have a breaking bore front. A careful design of numerical scheme is, of course,
necessary.

In order to ensure a stable computation, the grid lengthx and the time stept should

satisfy the following CFL condition.

x/t(2gh)1/2

(10)

It is not easy to solve the tsunami front that runs up and down the land with the shallow-water
theory expressed in the Eulerian description.

We need a moving boundary assumption.

In the

leap-frog numerical scheme, grid points are alternatively located for velocity and water level. Assume

- 10 -

that the water level is already computed at a computation cell. Then, compare the water level with the
bottom height of the next landward cell. If the water level is higher than the latter, the water may flow
into the landward cell. Iwasaki and Mano (1979) assume that the line connecting the water level and the
bottom height gives the surface slope to the first-order approximation. Hibberd and Peregrine (1979)
proposed more accurate method that required repeated computation. Aida (1977) and Houston - Butler
(1979) used weir formula to determine the discharge into the landward dry cell. For the practical
application, the Iwasaki and Mano assumption is widely used, because of its simplicity.
When a difference equation is solved, we have to control numerical errors that depend upon the
scheme used in the computation. Under the condition that the shallow-water equations are discretized
with the leap-frog scheme, one local tsunami wavelength should be covered by more than 20 grid points.
Otherwise, the computed tsunami height experiences numerical decay (Shuto et al., 1986).

Imamura and

Goto (1988) discussed more thoroughly the possible numerical errors in relation to local wave length.
If the Iwasaki-Mano boundary condition is used, the following condition should be satisfied at the
front.

x/gT2<410-4

(11)

where is the slope inclination and T is the wave period (Goto and Shuto, 1983).
Fujima and Shigemura (2000) proposed a way to control the numerical error in relation to the local
bottom topography.

7.3 2D/3D Hybrid Simulation

Under some special topographical condition, the long wave equation does not work well. In 1993,
the Hokkaido Nansei-Oki Earthquake Tsunami left the highest run-up in a small valley at Monai in the
Okushiri Island. The valley mouth 50 m wide was open at the sea shore and the tsunami trace at the
entrance of valley was about 23 m high. About 50 m apart from the entrance, the highest run-up of 31.7
m was measured on the valley bottom, suggesting that the vertical acceleration should not be neglected.
This accelerates the introduction of 2D/3D hybrid simulation; two-dimensional long wave equations for
wide areas and three dimensional original wave equation for the area where vertical acceleration shold be
fully taken into consideration. If applied to the case of Okushiri Island, the two-dimensional simulation
outside the valley is continued to the three-dimensional simulation in the valley in the neighborhood of
the valley entrance (Yoneyama et al., 2002).
The method of Fujima et al. (2002) and the STOC code developed by Tomita et al. (2004) are typical
2D/3D hybrid simulation.

Not only in run-up simulation but also in computation to simulate the

deformation and destruction of structure, the idea of 2D/3D is now used enthusiastically to develop new
numerical methods such as SPH (Smoothed Particle Hydrodynamics), DEM (Discrete Element Method),
and so on.
7.4 Aida Measures, K and
There are five sources of error in tsunami simulation.
in Section 4, there is no way to measure it.

The first is the initial profile. As discussed

The second source is the bathymetry that governs the

refraction and concentration of tsunamis. An accurate bathymetry in the sea shallower than 200 m is

- 11 -

essentially important. The third is the selection of basic differential equations. If the original equations
are used without any approximation, the CPU time is too long for practical application.

An

approximation, in which some terms are neglected or simplified, does not express the real phenomena.
In the difference equations, we have to expect inevitable truncation errors. Accuracy of measured data
that are used to verify the numerical results is not always reliable.
reflect the splash that cannot be simulated.

Tsunami trace heights sometimes

Tide records give smaller amplitudes than the actual tsunami

(Satake et al., 1988).

Aidas K and are used to judge whether or not a simulation gives satisfactory run-up heights
(Aida, 1978).
(12)
(13)
The measure K is a geometric mean of Ki, the ratio of the
measured run-up height Ri to the computed Hi,. The corresponding standard deviation is.

Based upon

rich experience of simulation in Japan, if K falls between 1.2 and 0.8 andis less than 1.4, the simulation
is judged satisfactorily carried out for practical use.

8. Mechanism and Damage Assessment

8.1 Damage to Houses and Buildings
(1) Damage to a Village Inundated
Hatori (1984) defined the damage percentage of houses RHD in a given village as follows.
RHD = (a + 0.5b)/(a + b +c)

(14)

where a is the number of houses washed away and completely destroyed, b is the number of houses
partially damaged, and c is the number of houses only flooded without structural damage.

Degree of

damage is judged by on-site inspection. Number of data thus obtained is of the order of several tens at
most.

He tried to find a relation between RHD and tsunami height. If numerical results are available,

RHD can be explained in terms of drag forces that is

proportional to current velocity (for example; Shuto, 1993).
Koshimura et al. (2009) developed Hatoris damage
percentage to fragility functions that treat tremendous
number of houses.

From a comparison of the pre-and

post-tsunami IKONOS satellite imageries for Banda Ache

hit by the 2004 Indian Ocean Tsunami, they counted the
16,474 destroyed houses, the roofs of which disappeared,
and the 32,436 survived houses of remaining roofs.
Combining the damage probability and inundation depth
(or, current velocity or hydrodynamic force), they obtained
the fragility functions as shown in Fig. 7

- 12 -

Figure 7: Example of fragility function.

(2) Damage to Individual Building

Degree of damage depends upon the structure of houses and the tsunami force. Data in the past are
given in terms of the inundation depth (Shuto, 1993).

Roughly speaking, a Japanese wooden house can

withstands if the inundation depth is smaller than 1 m and is completely washed away if the inundation
depth is larger than 2 m.

All the reinforced concrete buildings in the past example could withstand

tsunami forces even if the inundation depth was 5 m, except one example, the Scotch Cap Lighthouse that
was completely washed away by the 1946 Aleutian tsunami of 20 m inundation depth.
Iizuka and Matsutomi (2000) expressed quantitatively damage conditions in terms of inundation depth,
current velocity and/or hydrodynamic force. A wooden house will be destroyed if the inundation depth
is over 2 m, or if current velocity is over 4.9 m/s, or if hydrodynamic force is over 27 kN/m.
Koshimura et al. (2000) obtained that the structures were significantly vulnerable when the local
inundation depth exceeds 2 or 3 m, the current velocity exceeds 2.5 m/s or hydrodynamic load on a
structure exceeds 5 kN/m.

8.2 Impact of Waves and Floating Materials

(1) Wave pressure to vertical wall on land
Asakura et al. (2000) carried out hydraulic
experiments of tsunami force on buildings due to
tsunami run-up front without breaking.

Two types of

front, with and without formation of solitons, were

examined.

A run-up front without solitons gives

hydrostatic wave pressure distribution pH(z)

as

follows, where hc is the maximum inundation depth

measured on the seaward wall of building.

Figure 8: Pressure distribution without and

with solitons.

PH(z) /g = (3hc z)

(15)

For a run-up front with solitons, the lower part of the pressure distribution is modified as shown in
Fig.8 and the total pressure increases by 20 %. Including the two conditions, pressure distribution is
expressed as follows.
PH(z)/g = max(5.4 hc z, 3hc z)

(16)

Arikawa et al. (2006) carried out large-scale hydraulic experiments and showed that the equation
above was applicable for Froude number smaller than unity.

For Froude number that exceeds unity, a

tsunami can break just in front of wall to generate larger impact force.

The pressure distribution is not

hydrostatic but has a peak value at the inundation height.

The total force FD is defined by Iizuka and Matsutomi (2000) as follows,
FD = 0.61gCDhc2B

(17)

where CD (=1.1~2.0) is the drag coefficient and B is the breadth of building. Large range of variation in
the value of CD means that the shape of tsunami hitting the wall is important factor to generate whether
impact force or hydrostatic pressure.

- 13 -

(2) Force caused by breaking bore

The first hydraulic experiment of breaking bore in a large scale was carried out by Fukui et al.
(1962a).
The propagation velocity of the breaking front, c, was theoretically derived as follows.
c = [g(H + h)(H + 2h)/2(H + h H)]1/2

(18)

where H is the height of breaking bore, h is the water depth, is the total water depth andis the resistance
coefficient.
The impact total force, Pi, due to breaking bore was expressed in terms of this propagation velocity.
Pi = Kgc4/g2 H

(19)

where K is non-dimensional coefficient that takes 0.51 for vertical wall and tends to 0.33 as the wall slope
becomes gentle (Fukui et al, 1962b).
Run-up and force by breaking bores were experimentally studied by Mizutani and Imamura (2000).
They found three peaks in wave pressure on structures. The first is the dynamic wave pressure (D.W.P)
caused by impact of an incident bore. The second, the sustained wave pressure (S.W.P.), appears during
the high rise of water level because of continuous incidence of the bore. The third, the impact standing
wave pressure (I.S.W.P), is a result of impulsive collision between the incident and reflected bores.
Takahashi, Fujima and Asakura (2001) succeeded the numerical simulation of this phenomenon with a
method of numerical wave flume, CADMAS-SURF (SUper Roller Flume for Computer Aided Design of
MAritime Structure) (CDIT, 2001; Fujima, 2002). Figure 9 compares the time series of pressure along
the bottom surface and Fig. 10 compares the vertical distribution of wave pressure.

Figure 9: Computed and measured time series

Figure 10. Computed and measured wave pressure

of wave pressure.

distribution.

(3) Impact of floating materials

Figure 11: Pressure distribution with impact of lumber. Figure 12: Matsutomi diagram for lumber impact.

- 14 -

Hydraulic experiments (for example Matsutomi, 1993) showed the increase of force due to floating
materials.
Fig. 11 is an example of the vertical distribution of pressure when a lumber hits with the breaking
bore front. In addition to bore-induced water pressure, impact of a lumber acts locally but strongly.
Matsutomi (1999) provided a diagram (Fig.12) in terms of dimensionless impact and dimensionless
collision velocity to evaluate the impulsive force due to lumbers floated by breaking bores and surging
fronts.

In the figure, Fm is the impact, , D and L are the unit weight, diameter and length of lumber,

and VA0 is the approaching velocity just on collision.

After the 2004 Indian Ocean tsunami, similar studies but for different materials are being carried out,
for example, on ships and boats (Ikeya et al, 2006; Fujii et al, 2007), on cars (Anno et al., 2007), and on
containers (Kumagai et al., 2007) .
Transport of lumbers was first solved by Goto (1983).

Lumbers stored in timber yard on land begin

to be floated if the tsunami inundation depth exceeds their diameter and those in a timber basin when
tsunami force on timbers exceeds the strength of mooring wire.

Then, lumbers are transported by the

tsunami-induced current, scattering by diffusion effect.

8.3 Erosion due to Tsunami-Induced Current

(1) Damage to Coastal Road Embankment
Massive structures such as coastal road embankments made of soil are not destroyed by the strong
impact of tsunami front but are eroded and damaged by the water current induced by tsunamis.
type occurs when the tsunami height is lower than the crest of embankment.

The first

Stopped by a long coastal

embankment, the water concentrates to the openings such as underpass or bridge with increasing velocity.
Then, the neighborhood of the openings is soured.

The second is the case when a tsunami overflows the

structures and hit the rear slope and rear toe that are usually not protected with solid covers. The
overflowing tsunami is an unsteady flow that is sub-critical on the crest, super-critical on the rear slope
and returns to sub-critical after a hydraulic jump. Erosion process under this complicated flow is solved
by Fujii et al. (2009). They used CADMAS-SURF for flow computation combining with an erosion law
they established through hydraulic experiments.
(2) Toe of Quay Eroded by Backwash Water Falls
When a tsunami recedes after landing, the water falling from the top of quay wall directly hit the sea
bottom nearly exposed. The toe is scoured to lead to destruction of quay walls (Shuto, 2009).
Gotoh et al. (2002) used the moving-particle semi-implicit method (MPS) to solve the nap
formation of the falling water from the quay and the erosion of sand bottom at the toe.
Their result qualitatively explained the results of scouring obtained in large-scale hydraulic
experiments of Noguchi et al. (1997). However, there is no trial to explain some of damage
examples in the past.
(3) Tsunami-Induced Current in Narrow Waterways (Shuto, 2009)
The second case is the tsunami-induced current at narrow waterways in harbors or bays.

- 15 -

Strong

currents scour the sea bottom just at the toe of structures and destroy them.

A good example is found in

case of the gravity-type quay wall of the Konakano Fish market, Hachinohe Harbor, in Iwate Prefecture,
Japan.
The fish market was completed in August, 1959, one year before the 1960 Chilean Tsunami, near the
entrance of Hachinohe Industrial Harbor which was built by using the mouth of the Niida River. Near
the entrance of the water area 200 m wide and 2 km long, the Fish
harbor was very much influenced by the 1960 Chilean Tsunami.

Market was built. The whole

At the entrance of the harbor, the

maximum ebb flow velocity was estimated to be 13 m/s and the maximum flood flow velocity 8 m/s.
By this current, the Fish Market was affected and damaged.

Figure 13 shows a section of the market

before and after. The maximum amplitude of the 1960 Chilean Tsunami was about 6 m.
quay wall -3m deep was scored to -9 m.

The toe of the

In addition, the soil- and residual water- pressures from behind

during abnormally low water destroyed the basis of caissons and pushed them forward.
Five of 8 caissons, each of which was 10 m long, were overturned or subsided, as shown in Fig.13.
A person, who witnessed from the opposite side of harbor 200 m far, told that the quay wall collapsed
during an ebb tide from 06:31 a.m. to 07:03 a.m.

Figure 13: A section of the fish market damaged in 1960.

(4) Current Measurement
Different from the tsunami height, there are almost no measured data of tsunami-induced current.
Takahashi et al. (1991) used aero photographs taken for the Kesen-numa Bay in case of the 1960
Chilean Tsunami.

From the movement of floating materials and the Cameron effect, they determined

the current velocity distribution and compared with the results of their numerical simulation as in Fig. 15.
White circles were obtained from the numerical simulation for the sea bottom bathymetry before the
tsunami and black triangles are for that after the tsunami.

There are big differences.

The computed current velocity are less than half the measured, although their computation simulated very
well a tide record in the bay.
Nagai et al. (2004) concluded the generation of edge waves along an arch-shaped coast of Hokkaido
in case of the 2003 Tokachi-Off Earthquake Tsunami, using the records obtained with super-sonic current
meters installed by NOWPHAS (Nationwide Ocean Wave Information network for Ports and HAbourS) .
Use of videos was begun with the 2004 Indian Ocean Tsunami.

- 16 -

Fritz et al. (2006) analyzed an

eyewitness video record with PIV (planar particle image velocimetry) and obtained time series of flow
velocity at two locations which were within the range of 2 to 5 m/s.

Figure 14: Plan of Kesen-numa Bay

Figure 15: Computed and measured velocity

8.4 Tsunami Control Forest

There are discordant opinions about the effectiveness of a forest along a shoreline on the reduction of
tsunami energy.
Affirmative views assert that a forest is effective because; 1) it stops driftwood and other floating
materials, 2) it reduces water flow velocity and inundation water depth, 3) it provides a life-saving means
by catching persons carried off by tsunamis, and 4) it collects wind-blown sands and raises dunes, which
act as a natural barrier against tsunamis. A representative negative opinion is that a forest may be
ineffective against a huge tsunami, and at worst, trees themselves could become destructive forces to
houses if cut down by the tsunami.

Shuto (1987) collected forty-five examples for five huge tsunamis in

Japan and gave Fig. 12 in case of pine trees.

The ordinate is the summed tree
diameter nd, where n is the average number
of trees along the direction of water flow in
a rectangle with a frontage of unit length of
shoreline and a depth equal to the width of
forest, and d is the diameter breast high.
In region A, no trees are damaged.

The

number of trees is not enough to reduce

tsunami energy but is sufficient only to stop
boats and drift-wood. In Region B, trees
may be damaged. In Sub-region C-1, if there
is dense undergrowth in the forest, the
reduction of tsunami energy as well as the
stoppage of floating materials is expected.
In Sub-region D-1, the forest is thick enough

Fig.16 Shuto diagram for the effect of coastal forest.

- 17 -

and a similar effect to Sub-region C-1 is expected, even without undergrowth.

Harada and Imamura (2003) opened the way for quantitative evaluation of tsunami control forest,
using the shallow-water equations in which the effect of trees was expressed by Morison formula with
drag and inertia forces.

Experiences in 2004 are attracting many researchers to this topic, for example,

Tanaka et al. (2006) and Tanimoto et al. (2007).

9. Research Topics in the Near Future

Our knowledge of tsunami, tsunami damage and countermeasures is still limited. Among many
research topics, the present author would like to put emphasis on the following three subjects.
The first is related with tsunami-induced currents. Current velocity measured for real tsunamis is too
poor, in number and in quality. This is a big hindrance for further development of numerical simulation
and design of countermeasure structures. In addition to increase number of instruments in the field,
large scale hydraulic experiments are necessary. With these data, such defect as discussed in 8.3.(3) will
be improved. The simulation technique thus improved will become a powerful tool in the tsunami
archeology too, with which tsunamis not recorded on documents are found from tsunami deposit.
The second is the further development of numerical technique such as CADMAS-SURF. We need
large-scale hydraulic experiments to supply verification data for numerical simulation. An extension of
CADMAS-SURF to the three-dimensional space may become a powerful means to solve practical
problems.
The third is the use of CG animation in public education. The last way to save lives is an early
evacuation. Human beings, however, are optimistic.

Even in an emergency, we are likely to consider,

Im OK, and do not take necessary action. Due to this normalcy bias, many people have lost their
lives.

In order to break this preconception, CG animation works well. Katada et al (2004) developed a

kind of dynamic hazard map, and used in public

education.

Visit

his home

page

http://dsel.ce.gunma-u.ac.jp/, you can down load some of his results. Another use of CG technique is
the virtual realty, in which visitors can have a pseudo-experience of tsunami risk.

10. Concluding Remarks

Tsunami research started in Japan, in 1933 when the Showa Great Sanriku Tsunami hit.
Until 1960, it progress was slow but the theoretical approach was carried out by few pioneers, as the
tsunami science. After the 1960 Chilean tsunami, the tsunami engineering appeared to deal with the
actual risks. At that time, there were two countries where tsunami research was earnestly carried out,
Japan and USA.
In 1970s, tsunami researchers in USA, except for forecasting people, moved to other field such as
ocean engineering. In 1968, a local tsunami hit the Pacific coast of Japan and was nearly perfectly
prevented by coastal structures built after the Chilean tsunami.
tsunami research in Japan, too.

Then, it was not easy to continue

Under these conditions, supported by the electronic high-speed

computer and the Mansinha-Smylie method to determine the tsunami initial profile, both appeared in
1970s, tsunami science began a big step of progress with quite a few tsunami researchers.
In 1983 when a huge tsunami hit the Japan Sea coast and gave tremendous damages, USA people

- 18 -

were awoken and number of tsunami researcher in Japan began to increase, in the field of science and
engineering.
In September, 1992, the first international tsunami survey team (ITST) was formed to study the
Nicaraguan tsunami.

Then, in December of the same year, another tsunami hit the Flores Island,

Indonesia, ITST was also organized. In July, 1993, a huge tsunami occurred in the Japan Sea. This
frequent occurrence of tsunami disaster worked to increase number of tsunami researchers in the world.
Knowing a fact that the 1993 tsunami easily overflowed the man-made structures, it was confirmed that
human action was vitally important in an emergency.

This accelerated the participation of the social

scientists, with an emphasis on public education.

Each time when a tsunami occurs, tsunami research shows a progress and residents renew their
awareness. However, with the elapse of time, generation changes and precious experience will be lost.
Then, a new tsunami may ask new victims again. We, not only coastal residents but also tsunami
researchers, should resist this tendency, the decay of memory and prevention technique.

References
Aida, I. (1977): Numerical Experiments for Inundation of Tsunamis, Susaki and Usa, in the Kochi
Prefecture, Bulletin of Earthquake Research Institute, University of Tokyo, Vol.52, pp.441-460 (in
Japanese).
Aida, I. (1978): Rreliability of a Tsunami Source Model Derived from Fault Parameter, Journal of Physics
of Earth, Vol.26, pp.57-73.
Anno, K., T. Nishihata and Y. Moriya (2007): Numerical and Experimental Study on Drift Due to
Tsunami for Several Drifting Bodies, Annual Journal of Coastal Engineering, JSCE, Vol.54,
pp.866-870. (in Japanese)
Arikawa, T. et al. (2006): Large Model Test on Surge front Tsunami Force, Annual Journal of Coastal
Engineering, JSCE, Vol.53, pp.797-800.
Asakura, R. et al. (2000): An Experimental Study on Wave Force Due to Tsunamis That Overflow the
Upright Seawall, Proc. Coastal Engineering, JSCE, 47, pp.911-955 (in Japanese).
Carrier, G. F. and H. P. Greenspan (1958): Water Waves of Finite Amplitude on a Sloping Beach, Journal
of Fluid Mechanics, Vol.4, Part 1, pp.97-107.
CDIT (Coastal Development Institute of Technology) (2001): CADMAS-SURF, Research and
Development of Numerical Wave Channel, 296p (in Japanese) .
Dingemans, M. W. (1973): Water Waves Over an Uneven Bottom A Discussion of Long-Wave
Equations, Report R 729-II, Delft Hydraulic Laboratory, Delft, The Netherlands.
Fritz, H. M., J. Borrero, C. E. Synolakis and J. Yoo (2006): 2004 Indian Ocean Tsunami Flow Velocity
Measurements from Survivor Videos, Geophysical Research Letters, Vol.33, L24605,
doi:10.1029/2006GL026784, 5p.
Fujii, H., S. Hotta and N. Shuto (2000): Numerical Simulation of Damage to Soil Embankment Caused
by Overflowing Tsunamis, Journal of Disaster Research, Vol.4, No.6 (in press).
Fujii, N., M. Ohmori, M. Takao, S. Kanayama and H. Ohtani (1998): On the Deformation of the Sea
Bottom Topography Due to Tsunami, Proc. Coastal Engineering, JSCE, 45, pp.376-380 (in

- 19 -

Japanese).
Fujii, N., et al. (2007): Experimental Study on Variability in Drifting Behavior Due to Tusnmi and Its
Evaluation Method, Annual Journal of Coastal Engineering, JSCE, Vol.54, pp.241-245 (in
Japanese).
Fujima, K. and T. Shigemura (2000): Determination Grid Size for Leap-frog Finite Difference Model to
Simulate Tsunamis around a Conical Island, Coastal Engineering Journal, Vol.42, No.2,
pp.197-210.
Fujima, K., K. Masamura and C. Goto (2002): Development of the 2D/3D Hybrid Model for Tsunami
Numerical Simulation, Coastal Engineering Journal, Vol.44, No.2, pp.373-397.
Fujima, K. (2002): Review: Development of Numerical Wave Flume CADMAS-SURF ((SUper Roller
Flume for Computer Aided Design of MAritime Structure), Proceedings of Coastal and Ocean
Engineering in Korea, Vol.13, pp.1-14.
Fukui, Y., H. Siraishi, M. Nakamura and Y. Sasaki (1962a): A Study on Tsunami; (1) Propagation Velocity
of Bore, Proceedings of Coastal Engineering in Japan, JSCE, Vol.9, pp.44-49 (in Japanese).
Fukui, Y., H. Siraishi, M. Nakamura and Y. Sasaki (1962b): A Study on Tsunami; (2) Effect of Bore on
Coastal Dikes, Proceedings of Coastal Engineering in Japan, JSCE, Vol.9, pp.50-54. (in Japanese)
Fukuyama, E. and K. Irikura (1986): Rupture Process of the 1983 Japan Sea (Akita-Oki) Earthquake
Using a Waveform Inversion Method, Bulletin of the Seismological Society of America, Vol.76,
No.6, pp.1623-1640.
Garces, M. et al. (2005): Deep Infrasound Radiated by the Sumatra Earthquake and Tsunami, EOS,
Transactions, American Geophysical Union, Vol.86, No.36, pp.317-320.
Goto, C (1983): Numerical Simulation of Timbers Transported by Tsunamis, Proceedings of Coastal
Engineering in Japan, JSCE, Vol.30, pp.594-597 (in Japanese).
Goto, C. and N. Shuto (1983): Numerical Simulation of Tsunami Propagations and Run-up, in K. Iida and
T. Iwasaki (eds), Tsunamis: Their Science and Engineering, Terra Science Publication Co.,
Tokyo/Reidel, Dordrecht, pp.439-451.
Gotoh, H., T. Sakai, M. Hayashi, K. Oda and H. Ikari (2002): Gridless Analysis of Toe Scouring of
Seawall Due to Tsunami Return Flow, Proc. Coastal Engineering, JSCE, 49, pp.46-50 (in Japanese).
Harada, K. and F. Imamura (2003): Utilization of Forest Control to Mitigate the Tsunami Damage and Its
Evaluation, Proc. Coastal Engineering, JSCE, Vol.50, pp.341-345 (in Japanese).
Hatori, T. (1984): On the Damage to Houses Due to Tsunamis, Bulletin of Earthquake Research Institute,
Vol.59, pp.433-439 (in Japanese).
Hibberd, S. and D.H. Peregrine (1979): Surf and Run-up on a Beach: A Uniform Bore, Journal of Fluid
Mechanics, Vol95, pp.323-345.
Houston, J. R. and H. L. Butler (1979): A numerical model for tsunami inundation, WES Technical
Report HL-79-2.
Ikeya, T. et al. (2006): Experimental and Analytical Study of Impulsive Forces by a Drifter Due to
Tsunami, Annual Journal of Coastal Engineering, JSCE, Vol.53, pp.276-280 (in Japanese).
Iizuka, H. and H. Matsutimi (2000): Estimation of Damage Caused by Tsunami Flow, Proceedings of
Coastal Engineering, JSCE, Vol.47, pp.381-385 (in Japanese).

- 20 -

Imamura, F. and C. Goto (1988): Truncation Error in Numerical Tsunami Simulation by the Finite
Difference Method, Coastal Engineering Journal, Vol.31, No.2, pp.245-263.
Imamura, F., N. Shuto and C. Goto (1990): Study on Numerical Simulation of The Transoceanic
Propagation of Tsunami, Part 2: Characteristics of Tsunami Propagating Over the Pacific Ocean,
Zisin, Journal of the Seismological Society of Japan, Vol.43, pp.389-402 (in Japanese).
Iwasaki, S. (1992): Tsunami Forecasting Technique Using Underwater Sounds and Air pressure waves,
Tsunami Engineering Technical Report, DCRD Tohoku University, No.9, pp.65-83. (in Japanese)
Iwasaki, T. and A. Mano (1979): Two-dimensional Numerical Simulation of Tsunami Run-ups in the
Eulerian Description, Proceedings of 26th Conference on Coastal Engineering, JSCE, pp.70-74 (in
Japanese).
Iwase, H., T. Mikami and C. Goto (1998): Practical Tsunami Numerical Simulation Model by Use of
Non-linear Dispersive Long Wave Theory, Journal of Hydraulic, Coastal and Environmental
Engineering, JSCE, No.600/II-44, pp.119-124 (in Japanese).
Iwase, H., T. Mikami, C. Goto and K. Fujima (2002): A Comparative Study of Nonlinear Dispersive Long
Wave Equations for Numerical Simulation of Tsunami, Journal of Hydraulic, Coastal and
Environmental Engineering, JSCE, No.705/II-59, pp. 101-114 (in Japanese).
Izumiya, T. and H. Nagaoka (1994): Generation of Acoustic Gravity Waves by Tsunamis and Possibility
of Tsunami Forecasting, Proceedings of Coastal Engineering, JSCE, Vol.41, pp. 241-245 (in
Japanese).
Kajiura, K. (1970): Tsunami Source, Energy and the Directivity of Wave Radiation, Bulletin ERI, Vol.48,
pp. 835-869.
Katada, K., N. Kuwasawa and H. Yeh (2004): Development of Tsunami Comprehensive Scenario
Simulator for Disaster Education and Risk Management, National Geophysical Research Institute,
Hazards2004, 75p.
Koike, N. et al. (2005): Tsunami Height Survey of the 2004 Off the Kii Peninsula Earthquake, Annual
Journal of Coastal Engineering, JSCE, Vol.52, pp. 1336-1340 (in Japanese).
Koshimura, S. (2002): Multiple Reflection of Tsunamis Incident on a Continental Slope, Journal of
Hydraulic, Coastal and Environmental Engineering, JSCE, No.705/II-59, pp. 151-160 (in
Japanese).
Koshimura, S., F. Imamura and N. Shuto (1999): Propagation of Obliquely Incident Tsunamis on a Slope,
Part II Characteristics of On-ridge Tsunamis, Coastal Engineering Journal, Vol.41, No.2, pp.
165-182.
Koshimura, S., T. Oie, H. Yanagisawa and F. Imamura (2009): Developing Fragility Functions for
Tsunami Damage Estimation Using Numerical Model and Post-tsunami Data from Banda Ache,
Indonesia, Coastal Engineering Journal, Vol.51, No. 243-273.
Kumagai, K., K. Oda and N. Fujii (2007): Simulation Model for Drift Behavior of Container Due to
Tsunami and Collision Force, Annual Journal of Coastal Engineering, JSCE, Vol.54, pp. 236-240
(in Japanese).
Lander, J. F. (1996): Tsunamis Affecting Alaska 1737-1996, NGDC Key to Geophysical Research,
Documentation No.31, 195 p.

- 21 -

Madsen, P. A. and O. R. Sorensen (1992): A New Form of the Boussinesq Equations with Improved
Linear Dispersion Characteristics, Part 2; A Slowly-Varying Bathymetry, Coastal Engineering,
Vol.18, pp.183-204.
Manshinha, L. and D. E. Smylie (1971): The Displacement Fields of Inclined Faults, Bulletin of
Seismological Society of America, Vol.61, pp.1433-1440.
Matsutomi, H. (1993): Laboratory Study on Maximum Impulsive Force of Timbers Drifted by Bore-like
Tsunamis, Journal of Hydraulic, Coastal and Environmental Engineering, JSCE, No.467/II-23,
pp.19-28 (in Japanese).
Matsutomi, H. (1999): A Practical Formula for Estimating Impulsive Force Due to Driftwoods and
Variation Features of the Impulsive Force, Journal of Hydraulic, Coastal and Environmental
Engineering, JSCE, No.621/II-47, pp. 111-127 (in Japanese).
Mei, C. C. and B. LeMehaute (1966): Note on Equations of Long Waves Over an Uneven Bottom,
Journal of Geophysical Research, Vol. 71, pp. 393-400.
Mizutani, M. and F. Imamura (2000): Experiments on Bore Wave Pressure on Structures, Proceeding of
Coastal Engineering, Vol. 47, pp. 946-950 (in Japanese).
Mofjeld, H. O., V. V. Titov, F. I. Gonzalez, and J. C. Newman (2001): Tsunami Scattering Provinces in the
Pacific Ocean, Geophysical Research Letters, Vol. 28, No. 2, pp. 335-337.
Nagai, N., H. Ogawa, K. Nukada and M. Kudaka (2004): Characteristics of the Observed 2003
Tokachi-Off Earthquake Tsunami Profile, Annual Journal of Coastal Engineering, JSCE, Vol. 51, pp.
276-280 (in Japanese).
NHK (1983): A Huge Tsunami Witnessed Records of the Nihon-Kai Chubu Earthquake-, Video or DVD.
Noguchi, K., S. Sato and S. Tanaka (1997): Large-scale Experiments on Tsunami Overtopping and Bed
Scour around Coastal Revetment, Proc. Coastal Engineering, JSCE, Vol. 44, pp. 296-300 (in
Japanese).
Okada, Y. (1985): Surface Deformation Due to Shear and Tensile Faults in a Half-space, Bulletin of the
Seismological Society of America, Vol. 75, pp. 1135-1154.
Okal, E. A., J. Talandier and D. Reymond (2007): Quantification of Hydrophone Records of the 2004
Sumatra Tsunami, Pure and applied Geophysics, Vol. 164, pp. 309-323.
Omachi, T., H. Tukiyama and H. Matsumoto (1999): Evaluation of Tsunami Simulation Considering
Dynamic Crustal Displacement Accompanying a Fault Movement, Proceedings of Coastal
Engineering, JSCE, Vol. 46, pp. 321-325 (in Japanese).
Peregrine, D. H. (1967): Long Waves on a Beach, Journal of Fluid Mechanics, Vol. 71, Part 4, pp.
815-827.
Plafker, G. (1965): Tectonic Deformation Associated with the 1964 Alaska Earthquake, Science, Vol. 148,
pp. 1675-1687.
Satake, K., M. Okada and Ku. Abe (1988): Tide Gauge Response to Tsunamis: Measurements at 40 Tide
Gauge Stations in Japan, Journal of Marine Research, Vol. 46, pp. 557-571.
Satake, K. (1989): Inversion of Tsunami Waveforms for the Estimation of Heterogeneous Fault Motion of
Large Submarine Earthquakes: 1968 Tokachi-oki and 1983 Japan Sea earthquakes, Journal of
Geophysical Research, Vol. 98, pp. 4553-4565.

- 22 -

Shuto, N. (1967): Run-up of Long Waves on a Sloping Beach, Coastal Engineering in Japan, JSCE, Vol.
10, pp. 23-38.
Shuto, N. (1987): The Effectiveness and Limit of Tsunami Control Forests, Coastal Engineering in Japan,
Vol.30, No. 1, pp. 143-153.
Shuto, N. (1993): Tsunami Intensity and Disasters, Advances in Natural and Technological Hazards
Research, Vol.1, pp197-216.
Shuto, N. (2009): Damage to Coastal Structures by Tsunami-Induced Currents in the Past, Journal of
Disaster Research, Vol. 4, No. 6 (in press).
Simkin, T. and R, S, Fiske, (1983): Krakatau 1883, Eruption and its Effects, Smithsonian Institution Press,
464p.
Takahashi, R. (1934): A Model Experiment on the Mechanism of Seismic Sea Wave Generation, Part 1,
Bulletin ERI, Supplementary Vol.1, Tokyo University, pp. 152-181.
Takahashi, T., F. Imamura and N. Shuto (1991): A study on Tsunami-induced Current and Sea Bottom
Deformation, Proc. Coastal Engineering, JSCE, Vol. 38, pp. 161-165 (in Japanese).
Takahashi, T., F. Imamura and N. Shuto (1992): Tsunami Simulation with Movable Sea Bottom, Proc.
Coastal Engineering, JSCE, Vol. 39, pp. 231-235 (in Japanese).
Takahashi, T., F. Imamura, and N. Shuto (1993): Examination of Applicability of Movable Bed Model by
Tsunamis, Proc. Coastal Engineering, JSCE, Vol. 40, pp. 171-175 (in Japanese).
Takahashi, T., T. Takahashi, N. Shuto, F. Imamura and M. Ortiz (1995): Source Models for the 1993
Hokkaido Nansei-Oki Earthquake Tsunami, Pure and Applied Geophysics, Vol. 144, No. 3/4, pp.
747-767.
Takahashi, T., K. Fujima and R. Asakura (2001): An Application of Numerical Wave Flume to
Transformation of Bore, Proceedings of Civil Engineering in the Ocean, JSCE, Vol. 17, pp.
281-286 (in Japanese).
Tanimoto, K. (2007): Numerical Simulation of Tsunami Prevention by Coastal Forest with Several
Species of Tropical Tree, Annual Journal of Coastal Engineering, JSCE, Vol. 54, pp. 1381-1385 (in
Japanese).
Tomita, T., T. Kakinuma and A. Shimada (2004): Numerical Simulation of Tsunamis around and Behind
Breakwater using a 3D Model, Annual Journal of Coastal Engineering, JSCE, Vol. 51, pp. 296-300
(in Japanese).
Tanaka, N., T. Takemura, Y. Sasaki and M.I.M. Mowjood (2006): Breaking Condition of the
Representative Coastal Tree Species in Sri Lanka and the Effect of Vegetation on the Delay in
Tsunami Arrival Time, Annual Journal of Coastal Engineering, JSCE, Vol. 53, pp. 281-285 (in
Japanese).
Thucydides, (for example): History of The Peloponnesian War, Penguin Classics, p.247.
Tsuji, Y. (1977): A Study on the Scattering Wave Induced by Tsunami Passing Over a Sea Mount or Rise,
Ocean Science, Vol. 9, pp. 117-125 (in Japanese).
Uda, T. et al. (1988): Two-dimensional Deformation of Nonlinear Long Waves on a Beach, Report No.
2627, Public Works Research Institute, Ministry of Construction (in Japanese).
Yoneyama, N., M. Matsuyama and H. Tanaka (2002): Numerical Analysis for Locally High Runup of

- 23 -

1993 Hokkaido Nansei-Oki Seismic Tsunami, Journal of Hydraulic, Coastal and Environmental
Engineering, JSCE, No. 705/II-59, pp. 139-150 (in Japanese).

- 24 -

Proceedings of the 6th International

Workshop on Coastal Disaster Prevention,

Bangkok, Thailand, December 1-2, 2009

Tsunami Risk Assessment and the Planning and Implementation of

Strategic Mitigation Measures- Case Study City of Galle

Samantha HETTIARACHCHI
Department of Civil Engineering, University of Moratuwa, Moratuwa, Sri Lanka, sslh@civil.mrt.ac.lk

Saman SAMARAWICKRAMA
Department of Civil Engineering, University of Moratuwa, Moratuwa, Sri Lanka, sslh@civil.mrt.ac.lk

Nimal WIJERATNE
Faculty of Engineering, University of Ruhuna, Galle, Sri Lanka, nimal@ce.ruhuna.ac.lk

1. Risk - Components of Risk and Assessment

Planning post tsunami rehabilitation and conservation of the Sri Lankan coastline should ideally be
undertaken within a multi hazard coastal risk assessment framework giving due consideration to all the
coastal hazards. Even when risk assessments are undertaken only for the tsunami hazard it is important to
conduct such studies on a platform which can accommodate other coastal hazards. For risk assessment against
the tsunami hazard it is important to assess scientifically and establish the basis and criteria on which such an
exercise is undertaken. Planning based on observations arising from a single extreme event without
scientifically analyzing the true character of potential events, their impacts and future threats and risks should
be avoided.
Coastal communities all over the world are under severe pressure resulting from population growth in
coastal areas, human induced vulnerability, increases in frequency and magnitude of coastal hazards and
impacts of global climate change. These unprecedented changes are placing communities at increasing risk
from coastal hazards such as severe storms, tsunamis leading to coastal erosion, flooding and environmental
degradation. In this respect coastal community resilience is identified as the capacity to absorb and withstand
such impacts of hazards, emerge from disaster events and adapt efficiently to changing conditions.
The Indian Ocean Tsunami focused attention on globally on the severe impacts of tsunamis. It was also
recognized that coastal communities are increasingly at risk from a number of hazards which can be broadly
classified as Episodic and Chronic hazards. These hazards which may arise from natural phenomena or
human induces events have severe impact on coastal communities and eco-systems.

25

Episodic hazards include severe storms, earthquakes, tsunamis and oil spills all of which have limited
predictability and may result in major disasters. The communities should be made aware of these hazards,
their vulnerability and risks and should be educated on the importance of preparedness in responding to
potential disasters which usually require long term post event recovery efforts. Chronic conditions include
shoreline erosion, flooding, sedimentation, sea level rise and coastal environmental and resource degradation.
These condition which may result or increase from disasters arising from episodic hazards, relate to processes
which could be measured and monitored. They require long term planning measures and restoration efforts to
reduce risks.
Risk is usually expressed by the notation Risk = Hazard x Vulnerability. In this expression hazard includes
exposure. Risk represents the probability of harmful consequences or expected losses (in terms of deaths,
injuries, property, livelihoods, economic activity disrupted or environment affected) arising from interactions
between natural or human hazards and vulnerable conditions. Vulnerability can be broadly classified into
several components including, physical, human, socio- economic, functional and environmental vulnerability.
Hence vulnerability is dependent on several factors belonging to the said components, including population
density, building density and status, distance from the shoreline, elevation and evacuation time. Prior to the
Indian Ocean Tsunami (IOT), Sri Lanka had not adopted a planned approach towards preparedness in relation
to disasters, an aspect which is considered vital in saving lives. Hence the notation, Risk = Hazard x
Vulnerability x Deficiencies in Preparedness seems more appropriate. The additional term represents certain
measures and tasks the absence of which could reduce the loss of human lives and property in the specific
interval of time during which the event is taking place. This term is also commonly identified as the inverse of
Capacity. Hence there are many literature in which the notation, Risk = (Hazard x Vulnerability)/ Capacity is
used.
For detailed assessment of risk it is necessary to quantify the three main components of risk. However
quantifying all three terms is a challenging task in view of the wide range of diverse parameters associated
with the respective components of risk. In particular, there are no standard techniques for such assessment and
a number of methods have been used by researchers. These include qualitative methods, quantification based
on qualitative description and quantification based on detailed analysis of respective parameters. It is
therefore difficult to develop comprehensive risk assessment studies which capture the significance off all the
three components. However it is important that risk assessment studies are conducted within the framework
defined by the above formulae. This aspect has to be kept in mind when reviewing the outputs from studies on
risk assessment. The assessment of risk is an important element of coastal community resilience.
Communities must be made aware of the hazards, their exposure, vulnerability and be encouraged to address
issues on awareness, early warning, emergency planning, response and recovery and hazard mitigation.
Enhanced coastal community resilience enables populations at risk to live with risk with a greater degree of
confidence.

2. Increased Exposure of the City of Galle

Many coastal cities of Sri Lanka were severely affected by the Indian Ocean Tsunami due to the exposure
to the hazard. One of the principal coastal cities devastated was the historic port city of Galle (Figs. 1 and 2).

26

Incidentally the first recorded Tsunami to have affected Sri Lanka was on 27th August 1883, arising from the
eruption of the volcanic island of Krakatoa. On this occasion too, unusually high water levels followed by the
receding beach were observed in Galle around 1.30 pm. The water level fluctuations were not severe and
there was no inundation. The said time corresponds well with the tsunami travel time for tsunami waves
which would have been generated by the largest eruption of the volcano earlier in the morning. However, on
26th December 2004, Galle received the severe impact of Tsunami waves, their magnitude having increased
due to near-shore transformations. Galle is one of the many coastal cities around the world, which remains
heavily exposed to the tsunami hazard. Poorly constructed buildings and inadequate drainage contributes to
the vulnerability.
The tsunami waves, which reached the offshore waters of Galle were primarily diffracted waves,
diffraction taking place around the southern coast of Sri Lanka. In the context of Tsunamis the location of
Galle is extremely vulnerable. It lies besides a wide bay and a natural headland on which is located the
historic Galle Fort with very reflective vertical non-porous walls on all sides. Furthermore, there exists the
Dutch canal west of the headland, conveying water through the city centre. The waves in the vicinity of Galle,
which were increasing in height due to reduced water depths were further subjected to a series of near-shore
processes which increased their heights even further. The canal was a facilitator in conveying the massive
wave and associated flow towards the city centre.
In the vicinity of the headland on which the Galle Fort is located, the wave energy concentrates due to
refraction. These waves then reflected from the vertical solid walls of the Fort and moved around the
headland. Such walls reflect almost all the incident wave energy with very high wave heights at the wall itself.
There is hardly any dissipation. On the west of the headland the waves moved ferociously into the Dutch
Canal. On the east it moved along the bay. The wide bay in Galle further contributed to the increase in wave
height by modifying the shoaling process via reduced wave crest width to accommodate the bay shape. The
combined effect of this phenomenon and the wave coming around the eastern side of the Fort caused a
massive wave of destruction along the Marine Drive (see Fig. 3). It is certainly not surprising that many
survivors referred to a moving large black wall similar to that of the Galle Fort.
The city of Galle is therefore not only exposed to tsunami waves which will diffract around the southern
part of Sri Lanka it is even more exposed in the context of near-shore coastal processes which will further
increase wave heights. This aspect is identified as increased exposure within the risk assessment framework.
Figs. 1 and 2 give the testified tsunami wave heights and arrival times around Sri Lanka. Project location
Galle is indicated in Fig. 1.

3. Investigations for Risk Assessment Case Study

In order to safeguard lives and protect infrastructure a Risk Assessment Case Study was undertaken for the
City of Galle. One of the main objectives was to develop a Tsunami Hazard Map and an Evacuation Plan for
the City of Galle coastal area. Field surveys were carried out to collect data on:

Inundation height

Direction of first wave of tsunami

Possible evacuation paths & locations

27

34

9:20

9:00
5
58

<1
610

8:45

515

12

10:10

23

11:05

712
45
11

410
49 79

11:30

79

10:00

511

9:50

9:20
9:40

Figure 1: Testified Tsunami wave heights

Figure 2: Testified Tsunami arrival times

in meters.

(highest wave).

Bay increase of speed & height and circulation

Headland concentration of energy and

spreading around the headland
Figure 3: Galle Bay and Headland.

28

Area under the study was divided into 250m x250m grids and people living within the respective areas were
interviewed for all grids. The collected data were used to identify the

Inundated area

Inundation contours with wave direction and

Risk level of the area

Results of the study were also useful in identifying the

Safe areas and safe buildings

Evacuation routes and refuge areas

Proposed locations for fixing sign boards on evacuation routes

Fig. 4 gives the data collection points.

Figure 4: Data collection locations.

4. Numerical Modeling of Tsunamis

Numerical modeling of tsunami phenomena was carried out to obtain information on the coastal region of
Sri Lanka that could be affected by potential tsunamis. General coarse grid modeling was carried out for the
coastal region in the southern parts of the island and detailed fine grid modelling, including tsunami run-up
and inundation was carried out for the City of Galle. The results of the modelling was used for the preparation
of Hazard Map for Vulnerability for the City of Galle.
Generation and deepwater propagation of the tsunami waves were modeled using the AVI-NAMI model.
The module for co-seismic tsunami generation of AVI-NAMI uses the method developed by Okada (1985)
and the module for tsunami propagation solves Nonlinear Shallow Water Equations. ANUGA fluid dynamics
model based on a finite-volume method for solving Shallow Water Wave Equations was used for the

29

inundation modeling. In the ANUGA model the study area is represented by a mesh of triangular cells having
the flexibility to change the resolution of the mesh according to the area of importance. A major capability of
the model is that it can simulate the process of wetting and drying as water enters and leaves an area and
therefore suitable for simulating water flow onto a beach or dry land and around structures such as buildings.
High resolution near shore bathymetric data obtained for new Galle Port Development (2007) and high
resolution Topographic Data obtained after the 2004 Tsunami were used for study. (LIDA Surveys, 2005).
Broad scale deep water propagation Modeling was carried out for a number of source scenarios selected
from the Sunda/Java Trench (Table 1). Fault length of 500 km, a width of 150 km, Dip angle of 80, a Slip
angle of 1100 and a displacement of 40 m was used for the study.
Table 1: Source details and the maximum and minimum wave amplitudes from the propagation modeling
Longitude

Latitude

Strike Angle

Max. Amplitude

Min.

(m)

Amplitude (m)

Scenario 1

92.00' E

8.52' N

350'

2.015

-1.501

Scenario 2

94.26' E

3.09' N

329'

3.477

-2.391

Scenario 3

97.01' E

2.07' N

329'

1.419

-1.33

Scenario 4

97.60' E

-0.60' N

329'

2.608

-2.081

Based on the results of the deep water model, inundation modeling was carried out using the ANUGA
model. Modelling results give valuable information on the coastline of Galle that could be affected by
potential tsunamis. The model results are very useful for the preparation of Hazard maps. Fig. 5 gives the
inundation modeling results for 4 scenarios.

5. Countermeasures against the Tsunami Hazard - Classification and Planning

5.1 Classification of countermeasures
There are many countermeasures that could be adopted in coastal zone management when planning for a
tsunami and other coastal hazards that accompany high waves and high inundation. These include early
warning systems, regulatory interventions in the form of extending existing setback defense line and physical
interventions such as protection structures and utilizing the full potential of coastal ecosystems. These have to
be supplemented with public awareness on disaster preparedness, efficient evacuation procedures,
incorporating planned evacuation routes and structures that effectively integrate with the overall planning
process.
Countermeasures can therefore be broadly classified into two categories, namely, those which promote
successful evacuation from tsunamis and those which mitigate the impact of tsunami.

30

(a) Scenario 1

(b) Scenario 2

(c) Scenario 3

(d) Scenario 4

All inundation depths are in meters

Figure 5: Inundation Modelling for Galle- four selected scenarios.
5.1.1 Countermeasures that promote successful evacuation from tsunami
Countermeasures that promote successful evacuation from tsunami are listed below

1.

Early Warning Systems

2.

Public Warning Systems

3.

Hazard, Vulnerability and Risk Maps

4.

Set Back defense line

5.

Evacuation Routes and Structures

31

5.1.2 Countermeasures that mitigate the impact of tsunami

Countermeasures that mitigate the impact of tsunami
1.

The implementation of artificial measures for protection including tsunami breakwaters, dikes
and revetments

2.

The effective use of natural coastal ecosystems including Coral Reefs, Sand Dunes and Coastal
Vegetation (Mangrove Forests)

3.

Tsunami Resistant Buildings and Infrastructure

5.2 Planning Countermeasures via Policy and Management Options

It is important that post disaster planning should be undertaken in the context of overall coastal hazards one
of which remains Tsunamis, however remote the chances of an extreme event such as that of the 26th
December taking place. It is recognized that a Coastal Hazard Protection plan for a city which is an integral
part of an overall Coastal Zone Management Plan has to be based upon Policy and Management Options.
These options reflect the strategic approach for achieving long term stability in particular for sustaining
multiple uses of the coastal zone giving due consideration to the threats and risks of hazards.
Policy and Management options must be formulated on a sound scientific basis preferably to function within
the prevailing legal and institutional frameworks. However, if the need arises institutional improvements
should be affected and new laws should be imposed. In this process high priority should be given to
stakeholder participation. Extreme care has to be exercised when obtaining the active participation of
stakeholders who have witnessed and suffered heavily in terms of life, property and economic avenues from
one of the most sever natural disasters to have affected mankind. Most of them are yet to recover completely
from their traumatic experiences.
Policy Options identify possible courses of action on shoreline, as,
(1) Maintain existing defence line
(2) Setback defence line
(3) Retreat
(4) Advance
In order to implement the Policy Options various Management Options are considered provided they are
appropriate for the coastal classification. They are summarized as,
(1) Do nothing
(2) Reinstate to previous state
(3) Modify the existing design
(4) Develop new design
Once the Risk Assessment study is completed mitigation options will be developed within the framework
of Policy and Management Options with due consideration given to stakeholder consultations.
5.3 Classification of Physical Interventions (Artificial and Natural)
In the light of the discussion in Sections 5.1.2 and 5.2, mitigation by physical interventions is classified into
three types, depending on their location and function in protecting the coast. These interventions may be

32

achieved not only by artificial methods via Coastal Engineering Design but also by harnessing the full
potential of natural coastal ecosystems. The types of interventions and typical examples for each category are
listed below.
(i)

Reduce the impacts of tsunami waves prior to reaching the shoreline.

(eg. Tsunami Breakwaters, Coral Reefs)

(ii)

Protect the coastal zone by preventing the inland movement of tsunami waves.
(eg. Tsunami Dike, Sand Dunes)

(iii)

Mitigate the severe impacts of tsunami waves on entry to the shoreline.

(eg. Tsunami Dikes, Revetment, Mangrove Forests)

On many occasions both methods can be adopted in parallel to develop well-integrated hybrid solutions
satisfying environmental concerns.
5.4 Development of Guidelines for tsunami resistant buildings
The coast is an area of high economic activity and it is not possible to transfer all activities to areas that are
completely free from potential tsunami hazards. For some areas of the coast, safe evacuation areas may be too
far away for citizens to reach on foot thus necessitating vertical evacuation structures. Therefore there is a
need to develop Design Guidelines and Construction Manuals for tsunami resistant housing and infrastructure
for the benefit of the public and wider usage.
Given the background of discussion in Sections 5.1.2 and 5.2, two types of guidelines are required.
1) Overall Design Guidelines providing advice on location, layout, orientation, structural configuration,
geo-technical considerations and other considerations relating to good design practice.
2) Detailed Design Guidelines leading to hydraulic and structural loads, geo-technical issues and
detailed design information.
The Overall Design Guidelines could be developed from the experience gained from Damage Assessment
from different parts of the country and such assessment should be analyzed in the context of the hydraulic
regime which would have been generated by the tsunami at that location. Relevant information from other
countries that have been affected by tsunamis will also be very useful for this exercise. It is important that
Damage Assessment should cover infrastructure that was (i) Destroyed (ii) Damaged (iii) Survived (least
affected).

6. Integrating Mitigation with Development Projects- Port of Galle

In 2000, Japanese Port Consultants (JPC) developed a Master Plan for the development of the Port of Galle.
In view of environmental issues they recognized that the development should be restricted to a two berth
medium size harbour. In order to maintain healthy exchange of tidal flow for the well being of the coral reef
system, JPC in consultation with the environmental specialists incorporated an offshore detached breakwater,
which coincidentally has all the characteristics of an effective Tsunami Breakwater (see Fig. 6). It must be
admitted Tsunamis were not part of the agenda of the engineering and environmental terms at that stage.

33

Figure 6: Master Plan for the development of the Port of Galle.

Detail design of the port of Galle is being carried out by PCI, Japan. Tsunami modeling carried out
indicates a reduction in both Tsunami wave height and the speed due to the proposed offshore breakwater. A
hybrid approach using both artificial and natural methods seems very appropriate for this situation. These are
some of the Coastal Engineering mitigation measures, which can be considered in examining the options for
the protection of Galle. These measures will also be effective against potential coastal hazards that have a
greater probability of occurrence than a massive Tsunami wave.

7. Preparation of Hazard, Vulnerability, Risk and Evacuation Maps

Based on the investigations conducted in Sections 3,4 and 5 and Public Consultations via meetings with
affected people and questionnaire surveys, initial hazard, vulnerability, risk and evacuation maps were
prepared. These maps will be refined with further field investigations and modeling work. The said maps are
presented below in Figs. 7 and 8.

8. Concluding Remarks
The paper presented the approach adopted for risk assessment in the context of hazard, vulnerability and
capacity and provided a summary of the investigative studies carried for the tsunami risk assessment study for
the Port City of Galle. It discussed the mitigation measures with respect to coast conservation via bio-shields
and the development of a tsunami breakwater as part of strategic port development. It focused on the
development of building codes for new construction, enhancing the strength of existing buildings and their
field applications. Finally the paper discussed risk management measures in the form of community
preparedness, evacuation maps and evacuation zones and shelters.

34

Figure 7: Safe areas and safe buildings.

Figure 8: Evacuation Places and Routes.

References
Juan Carlos Villagran de Leon (2008): Rapid Assessment of Potential Impacts of Tsunami- Lessons from the
Port of Galle in Sri Lanka, United Nations University, Bonn, SOURCE No.9/2008.
Working Group on Risk Assessment of the Intergovernmental Coordination Group for the Establishment of
the Indian Ocean Tsunami Warning and Mitigation System (ICG/IOTWS) (2009): Guidelines on
Tsunami Risk assessment and Mitigation fro the Indian Ocean: Knowing your Tsunami Risk and What
to do about it., To be published as a UNESCO-IOC Manual and Guideline No 52 by UNESCO, Paris.

35

36

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

REAL TIME MONITORING FOR MEGA THRUST

EARTHQUAKES AND TSUNAMIS AROUND THE NANKAI
TROUGH SOUTHWESTERN JAPAN
-Towards to understanding mega thrust earthquakes and
disaster mitigationYoshiyuki KANEDA
Earthquake and Tsunami Research Project for Disaster Prevention, Japan Agency for Marine-Earth Science
and Technology (JAMSTEC), Yokosuka, Japan, kaneday@jamstec.go.jp

Katsuyoshi KAWAGUCHI
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
kawak@jamstec.go.jp

Eiichiro Araki
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
araki@jamstec.go.jp

Hiroyuki MATSUMOTO
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
hmatsumoto@jamstec.go.jp

Takeshi NAKAMURA
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
t_nakamura@jamstec.go.jp

Shinichiro KAMIYA
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
kamiya@jamstec.go.jp

Keisuke ARIYOSHI
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
Ariyoshi@jamstec.go.jp

Takane HORI
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
horit@jamstec.go.jp

Hide SAKAGUCHI
Institute for Research on Earth Evolution, JAMSTEC, Yokosuka, Japan, sakaguchih@jamstec.go.jp
- 37 -

Maddegedara LALITH
Institute for Research on Earth Evolution, JAMSTEC, Yokosuka, Japan, lalith@jamstec.go.jp

Toshitaka BABA
Earthquake and Tsunami Research Project for Disaster Prevention, JAMSTEC, Yokosuka, Japan,
babat@jamstec.go.jp

Abstract
Around the Nankai trough southwestern Japan, mega thrust earthquakes over M8 class such as
the Tokai, the Tonankai and the Nankai earthquake are occurring with intervals of 100-200
years.

In past two mega thrust earthquakes in 1944/1946 and 1854, first ruptures were

starting from the Tonankai seismogenic zone. Therefore, to elucidate the recurrences of the
Tonankai mega thrust earthquakes and mitigate geohazard, ocean floor network systems for the
real time monitoring of seismogenic zone are quite important. We are starting to develop the
Dense Ocean floor Network system for Earthquakes and Tsunamis (DONET) as Japanese
government funded MEXT project during 2006-2010.
Advantages of DONET system are as follows,
1) Early warning of mega thrust earthquakes and tsunamis,
2) Precise observation of crustal activities around seismogenic zones using a dense network
array with multi kinds of sensors, and
3) Data from ocean floor network system are quite important to improve recurrence cycle
simulation.
We are developing DONET system and will deploy around the Tonankai seismogenic zone.
DONET is equipped with multi kinds of sensors such as accelerometers, broad band
seismometers and precise pressure gauges etc.
There are three important concepts in DONET as follows,
1) Redundancy of system by the looped backbone cable,
2) Extendable and Replaceable system by nodes and brunch cables, and
3) Advanced technology and maintenance system by ROV / AUV applications.
The dense array data from DONET will be applied to the early warning earthquakes and
tsunamis. Furthermore, the precise data of crustal activities from DONET will improve the
recurrence cycle simulation model using data assimilations.
Furthermore, to understand the mega thrust recurrence system, expeditions by the drilling
vessel CHIKYU are carried out around the Tonankai seismogenic zone. Finally, for 4
dimensional real time monitoring of the Tonankai seismogenic zone, borehole observatories by
CHIKYU drilling will be connected to DONET system. These monitoring data will be
applied to improve the simulation model and the advanced early warning system.
Keywords: Tsunami; earthquake; geohazard; seismogenic

- 38 -

1. Introduction
In the Nankia trough, there are 3 mega thrust rupture zones such as the Tokai, Tonankai and Nankai
seismogenic zone (Fig. 1). In these rupture areas, mega thrust earthquakes over M8 are occurring with
intervals of 100-200years (Table1).

Figure 1: The Nankai seismogenic zone.

Table 1: The historical earthquakes around the Nankai trough. (Occurred Conjecture)

yeasr
684
887
1096/1099
1361
1498
1605
1707
1854
1944/1946

Nankai

Tonankai

Tokai

In 1854, 1944/46, actually, the initial rupture were starting from the Tonankai earthquake rupture zone
ahead of the Nankai earthquake rupture zone with intervals of 32 hours and 2 years in each event. In previous
simulation researches, the result of recurrence cycle simulation indicates the difference patterns and intervals
of mega-thrust earthquake recurrences in each cycle. And, according to results of simulations, first ruptures
are starting from the Tonankai earthquake rupture zone. These results are consisted with recent historical
earthquakes in 1854, 1944/46 around the Nankai trough.
The new research project for the evaluation of seismic linkage around the Nankai trough mega thrust
earthquakes is starting as MEXT project which is a kind of Japanese government.
- 39 -

Around the Nankai trough, especially, off Kii peninsula, this new research project including observations,
simulations and disaster mitigation researches, the advanced ocean floor network system (DONET) and
CHIKYU drilling as IODP (http://www.iodp.org/) Integrated Ocean Drilling Program are carrying out.

2. Previous Researches
We could have the detailed information of the 1944 Tonankai earthquake and1946 Nankai earthquake. The
1944 Tonankai and the 1946 Nankai earthquakes, each hypocenter was located off the Kii peninsula. So, the
imaged irregular structure such as a key structure at the segment boundary between the Tonankai and Nankai
earthquake rupture zone seems to be the controller of the Nankai Trough mega-thrust seismogenic zone
system (Fig. 1). Furthermore, the results of recent simulation study of mega-thrust earthquakes recurrence
cycles indicates that these irregular structures seem to act as a controller of recurrence cycle and pattern of
mega-thrust earthquakes in the Nankai trough. And in each simulation cycles, the first ruptures are starting
from the Tonankai earthquake rupture zone ahead of the Nankai earthquake rupture zone. These results are
consistent with the past two earthquakes in 1854, 1944/1946. (Table1)
Therefore, for real time monitoring of seismogenic zone, DONET will be deployed around the Tonankai
rupture zone.

3. The Dense Ocean Floor Network (DONET)

According to previous researches, the Tonankai seismogenic zone is important to understand the system
of Nankai trough mega thrust earthquake occurrences. Therefore, we proposed and have been starting to
deploy the dense ocean floor observatory network system equipped with multi kinds of sensors such as
seismometers, pressure gauges etc. around the Tonankai seismogenic zone (Fig. 2). Especially, precise multi
pressure gauges will be most useful sensor to not only detect tsunamis but also monitor ocean floor
deformations with long term observation. These ocean floor deformation data will be applied to the data
assimilation to improve recurrence cycle simulation model.

Figure 2: The outline of DONET.

This observatory system will be the one of most advanced scientific tools to understand the mega thrust
earthquakes around the Nankai trough.
This advanced dense ocean floor observatory network system has useful functions and purposes as
follows,
- 40 -

1) Redundancy, extension and advanced maintenance system using the looped cable system, junction boxes
and the ROV/AUV etc (Fig. 3)
We are carrying out the development of ROV for the 10km cable extension and heavy weight operation.
2) Multi kinds of sensors to observe broad band range phenomena such as long period tremors, low
frequency earthquakes and strong motion of mega thrust earthquakes over M8.(Fig. 4)
3) Speedy evaluation and notification for earthquakes and tsunamis.
This function is most important for disaster reduction /mitigation. JMA is starting the early warning
system, therefore, the ocean floor network data will contribute to this early warning system for disaster
mitigation. Especially, in the tsunami early warning, ocean floor network is absolutely necessary and
important (Fig. 5 (1), Fig. 5 (2)).
4) Provide observed data such as ocean floor deformation derived from pressure gauges to improve the
simulation and modeling researches about the mega thrust earthquakes. These ocean floor deformation
data are quiet necessary for the data assimilation to improve simulation models (Fig. 6).
5) Next Ocean floor network as DONET2 is planning to deploy around the western part of the Nankai
trough shown in Fig. 7
6) DONET2 has high voltage system to extend backbone
7) Cable and to equip with many nodes/sensors more. The outline of DONET2 system is shown in Fig. 8.
8) New research project for the estimate of seismic linkage around the Nankai trough mega thrust
earthquakes is starting from FY 2008 as 5 years project (Fig. 9).
This project is entrusted by MEXT project which is a kind of Japanese government.
DONET data will be applied to this new research project.

Figure 3: The concept of DONET.

Figure 4: The observing targets of DONET.

- 41 -

Figure 5 (1): The flow of speedy evaluation and notification for earthquakes and tsunamis.

Figure 5 (2): The tsunami early warning using DONET

Figure 6: The outline of data assimilation.

Figure 7: Future Plan of DONET

- 42 -

Figure 8: The outline of DONET2 system

Figure 9: New research project for the estimation of seismic linkage around the Nankai trough-Observation
Simulation and Disaster mitigationThis project is entrusted by MEXT project which is a kind of Japanese government.DONET data will be
applied to this new research project

4. Future Plan
We will deploy the advanced ocean floor network off Kii peninsula not only the Tonankai earthquake
rupture zone but the Nankai earthquake rupture zone as a local network system. We have to apply the data
from network to understand the detailed crustal activities, to improve simulation model, and to mitigate
disaster damages.
In the second step, we would like to develop and deploy the advanced ocean floor network as a regional
system and integrate ocean and land observatory network data using advanced ocean floor observatory
technologies.
In the next step, we would like to collaborate with international network systems such as NEPTUNE,
ESONET etc. to progress geosciences and contribute the early warning system for huge earthquakes and
tsunamis (Figure10).
Finally, we would like to enhance to the importance of practical database to improve structures and
simulation such as data assimilation.

- 43 -

Figure10: The collaboration of international ocean floor observatories.

References
Ariyoshi, K., T. Hori, J.-P. Ampuero, Y. Kaneda, T. Matsuzawa, R. Hino and A. Hasegawa: Influence of
Interaction between Small Asperities on Various Types of Slow Earthquakes in a 3-D Simulation for a
Subduction Plate Boundary, Gondwana Research.
Hori, T. (2006): Mechanisms of Separation of Rupture Area and Variation in Time Interval and Size of Great
Earthquakes along the Nankai Trough, Southwest Japan, J. Earth Simulator, Vol. 5, pp. 8-19.
Hori, T., N. Kato, K. Hirahara, T. Baba and Y. Kaneda (2004): A Numerical Simulation of Earthquake Cycles
along the Nankai Trough, Southwest Japan: Lateral Variation in Frictional Property Due to the Slab
Geometry Controls the Nucleation Position, Earth Planet. Sci. Lett., Vol. 228, pp. 215-226.
Kaneda, Y., DONET Group, AGU2007
Kaneda, Y., DONET Group, ICDP2007
Kaneda, Y., DONET Group, IUGG2007
Kaneda, Y., DONET Group, PICES2007
Kaneda, Y., DONET Group, SubFac-SEIZE2007
Kaneda, Y., DONET Group, WSEAS2008
Kaneda, Y., Ocean Observatory Summit
Kaneda, Y. (2008): Observational Research Group for the Next Nankai Trough Mega Thrust Earthquakes,
ASC2008
Kaneda, Y. (2009): Earthquake and Tsunami Research Pproject for Disaster Prevention, EGU2009
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, JPGU2009
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, ESONET NoE
Training 2009, Guest Speech.
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, IFREMER Mtg. 2009,
Guest Speech .
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, VISO WS 2009, Guest
Speech.
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, INGV Mtg. 2009,
Guest Speech.
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, Ocean Obs 09.
Kaneda, Y. (2009): Earthquake and Tsunami Research Project for Disaster Prevention, 2nd All Regions WS
- 44 -

ESONET 2009, Guest Speech.

Kaneda, Y., K .Hirahara and T. Furumura (2008): Research Groups for the Next Nankai Trough Mega Thrust
Earthquakes, AGU2008
Kaneda, Y., M. Ohori and T. Nakamura (2009): New Proposed Disaster Mitigation and Recovery for the Next
Nankai Trough Megathrust Earthquakes, Journal of Disaster Research, Vol. 14, No. 2, pp151-152.
Kaneda, Y. and S. Kodaira (2009): Structural Research on the Nankai Trough Using Reflections and
Refractions, Journal of Disaster Research, Vol. 4, No. 2, pp. 67-71.
Kodaira, S., N. Takahashi, A. Nakanishi, S. Miura and Y. Kaneda (2002): Subducted Seamount Imaged in the
Rupture Zone of the 1946 Nankaido Earthquake, Science, Vol. 289 pp. 104-106.
Kodaira, S., T. Iidaka, A. Kato, J.-O. Park, T. Iwasaki and Y. Kaneda (2004): High Pore Fluid Pressure May
Cause Silent Slip in the Nankai Trough, Science, Vol. 304, pp. 1295-1298.
Mitsui, N., T. Hori, S. Miyazaki, K. Hirahara and Y. Kaneda (2009): Constraining Inter plate Frictional
Parameters Using Limited Terms of Observation Data, JISHIN, Vol. 2, No. 61, pp. 149-159.
Nakamura, T., Y. Ishihara, Y. Yamanaka and Y. Kaneda: Source Process for Complex Fault System of the
2007 Chuetsu-oki, Niigata, Japan, Earthquake, Earth, Planets and Space, Vol. 61, pp. 273-278.
Park, J.-O, T. Tsuru, S. Kodaira, P Phil R. Cummins, and Yoshiyuki Kaneda (2002): Splay Fault Branching
along the Nankai Subduction Zone, Science, Vol. 297, pp. 1157-1160.

- 45 -

- 46 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

TSUNAMI WARNING SYSTEM IN JAPAN

Kenji NAKATA
Japan Meteorological Agency, Tokyo, Japan, ken-nakata@met.kishou.go.jp

Abstract
JMA has been operating the tsunami warning service for decades. Tsunami warning system
has recently acquired improvements such as utilization of GPS buoy data, change to Two
Main Centers system. JMAs international tsunami information for the northwest Pacific
area and Indian Ocean area provided as an international contribution is also described.
Keywords: Tsunami warning; GPS buoy; two Main centers; international information;

1. Introduction
Japan is located in the circum-Pacific tectonic zone where
seismic activity is extremely high in the world, and more than
ten percent of earthquakes in the world take place in and
around Japan. That leads Japan, surrounded by seas, to one of
the most tsunami-prone countries in the world, having
suffered from a lot of serious tsunami disasters. Under such
circumstances, the Japan Meteorological Agency (JMA)
established a national system of tsunami warning service in
1952. JMA has been operating the system since then, with
continuous improvement by introducing newest technologies
of the time.

2. Tsunami Warning Service

JMA continuously monitors seismic activity in and
around Japan to determine the hypocenter and the magnitude

Figure 1: 66 Regions for Tsunami

immediately after a large earthquake occurs. If an earthquake

Warning

occurs in the ocean area with tsunamigenic potential, JMA

conducts the tsunami warning operation using a database which contains estimated tsunami heights and
travel time for hundreds of thousands of scenarios which were calculated in advance. Tsunami
warnings/advisories are issued for 66 coastal regions (Fig. 1) with respective estimated maximum tsunami
heights and estimated tsunami arrival time. These warnings/advisories are provided for the national and
- 47 -

local authorities as well as for broadcasting stations. Governors of municipalities are responsible for
directing or instructing the residents to evacuate from tsunami hazardous areas.
In case of a large earthquake distant from Japan, JMA also determines the hypocenter and magnitude
using seismic data from global seismological observation network as well as domestic ones, and
exchanges information on the earthquake with PTWC and USGS. If there is a possibility of tsunami
generation, JMA conducts the domestic tsunami warning operation in the same manner and criteria as the
local tsunami procedure. If the location of the earthquake is in the northwest Pacific area on the Indian
Ocean, JMA also issues an international tsunami advisory or information.

3. Tsunami Monitoring and Re-Evaluation of Tsunami Warning

JMA monitors sea level changes at 160 tide
gauges located along coasts for the purpose of
detection and observation of tsunami. Besides
providing tsunami information with observed
heights and arrival time, if observed tsunami
heights exceed predicted heights, JMA re-evaluates
tsunami

warnings/advisories

and

updates

it

accordingly. JMA cancels warnings/advisories

when JMA concludes that the dangerous situation
is over, namely, when the observed heights of the
tsunami diminish and become lower.
In addition to coastal observation, JMA began

Figure 2: Tsunami detection system by GPS buoy

in July 2008 to use GPS buoys floating about 20 km offshore from the coasts for monitoring tsunami. (Fig.
2) This system is installed and operated by Ports and Harbours Bureau, Ministry of Land, Infrastructure,
Transport and Tourism (MLIT) and Port and Airport Research Institute (PARI). Early detection of
tsunami is expected by this equipment before tsunami arrives at the coasts. JMA uses 8 GPS buoys
around Japanese islands at present for detecting tsunami and issuing tsunami information. JMA has not
issued tsunami information including GPS buoys observation so far.

4. Change to Two Main Centers Tsunami Warning System

JMA updated earthquake processing systems at Osaka District Observatory on March 2009, and at
JMA headquarters on 1st Oct. 2009. This accompanied a change of the Tsunami Warning System to Two
Main Centers system. Osaka became another main center of tsunami warning. (Fig. 3)
Before the update, JMA had one main center and five regional centers for tsunami warning. The main
center at Tokyo covered all over Japan and five regional centers covered the responsible regions. A new
main center at Osaka has all the data collected from the seismic stations and can do all the processing
from the determination of hypocenter to issuance of tsunami warning, which are the same function as
Tokyo has. (International tsunami information service remains to be operated only in Tokyo.) The new
system was realized by the progress of the telecommunication link.
This Two Main Centers system has enabled one main center to issue the tsunami warning smoothly and
speedy when another main center loses its center function by damages from a large earthquake or other
- 48 -

accidents. It is also expected that costs of update and maintenance will become smaller.

Main center

Osak

Regional

Tokyo
Main center
Regional
Figure 3: Updated structure of warning centers of JMA

5. International Tsunami Advisory/Information by JMA and Indian Ocean

Tsunami Exercise (IOWave09)
JMA issues two kinds of international tsunami advisory/information, NWPTA (Northwest Pacific
Tsunami Advisory) for the northwest Pacific area and TWI (Tsunami Watch Information) for the Indian
Ocean. NWPTA started to be provided in 2005 as a regional service under the framework of the
Intergovernmental Coordination Group for the Pacific Ocean Tsunami Warning and Mitigation System
(ICG/PTWS). TWI started to be provided in 2005 under the framework of the Intergovernmental
Coordination Group for the Indian Ocean Tsunami Warning and Mitigation System (ICG/IOTWS). TWI
service will continue until tsunami warning system will be established by the countries in the Indian
Ocean region and be in full operation.
When an earthquake of magnitude 6.5 or greater occurs in the respective coverage area, NWPTA or
TWI is provided to relevant countries in about 20 to 30 minutes. NWPTA and TWI contain earthquake
parameters, evaluation of tsunamigenic potential and estimated tsunami arrival times or travel times.
Besides, information with regard to estimated tsunami heights is also provided in NWPTA. When a
tsunami is actually observed at tidal stations, information on tsunami observations is subsequently
provided.
Since it is very important to ensure transmission of the information, JMA regularly conducts
communication test for both of NWPTA and TWI. Having the same function as one of the objectives, the
first Indian Ocean Tsunami Exercise (IOWave09) under the framework of the ICG/IOTWS was
conducted on 14 Oct. 2009. While 18 countries in the Indian Ocean region participated in the exercise,
JMA as well as PTWC took important roles to support drawing up the exercise scenario and to send the
information for the exercise. The exercise ended successfully.

- 49 -

- 50 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

EXPERIMENTAL STUDY ON TSUNAMI POWER

Taro ARIKAWA
Tsunami Research Center, Port and Airport Res. Inst., Yokosuka, Japan, arikawa@pari.go.jp
Abstract
This report describes surge front tsunami force using physical model experiments. Wooden
house walls and concrete walls 6 cm and 10.0 cm thick in a flume to investigate the
land-structure destruction mechanism of tsunami inundation force broke the instant the 2.5m
tsunami hit walls.The difference of the failure mode of concrete wall due to impulsive tsunami
force was clarified. If the strength of wall is small, then the bending failure or punching shear
failure is occurred. When the strength of wall is increasing, the failure mode is shifting local
failure to whole destruction.
Keywords: Tsunami; Large scale experiments;

1. Introduction
The December 26, 2004 Indian Ocean Tsunami killing over 200,000 and damaging innumerable land
structures reminded the world again of the terrific power of such tsunamis.
Regarding tsunami force hitting land structures called surge front tsunami force (Matsutomi, 1991), Iizuka
and Matsutomi (2000) proposed a logic-based empirical formula stating that surge front force was
proportional to the square of flow velocity. Asakura et al. (2000) suggested that force has a triangular
distribution equivalent to hydrostatic load distributed over a vertical distance equal to three times the
inundation height.
Photo 1 shows damage to a house in Galle, Sri Lanka hit by a tsunami with an average inundation height
of 2.5m, further shown by the tsunami trace to have peaked at 3.8m. The house wall was made of mortar and
brick. Key tsunami parameters are unknown, so basic experiments on surge front tsunami force were
conducted.

Photo 1: Housing damage in Galle City, Sri Lanka, photographed on Apr. 21, 2005.
1

- 51 -

2. Surge Front Tsunami Force

2.1 Classification
Surge front tsunami force is qualitatively divided based on the speed and wave profile into the types
shown in Fig. 1:
(i) Type 1: Overflow, associated with slow speed.
(ii) Type 2: Bore, associated with quick flow after breaking or resulting from the supercritical flow.
(iii) Type 3: Breaking, associated with the tsunami breaking in front of a land structure, usually near a
coastline with a steep seabed slope. It occurs when the wave breaks just in front of a structure. Then very
large impulsive pressure is occurred. The maximum value of pressure is depend on not only tsunami height
but also the wave profile.
For a given tsunami inundation height, the maximum force is largest for a breaking type impact as
opposed to an overflow type of inundation. However, it is not clear how large the force is and where a
boundary line of type is.

Figure 1: Type of surge front force

2.2 Definitions
Fig. 2 shows a typical time history of the surge front tsunami pressure. Since the tsunami period
usually exceeds several minutes, the main tsunami pressure is the same as a continuously acting steady flow,
whose pressure is called sustained pressure. Leading tsunami pressure, called bore pressure, depends
greatly on the wave profile and may exceed sustained pressure, depending on the case.

Figure 2: Typical tsunami pressure time history

Asakura et al. (2000) proposed the following empirical formula based on research focusing on sustained

- 522 -

pressure and its vertical distribution:

for 0 z max 3
p z g max 31 z 3 max
p is pressure, z the vertical axis,

max

(1)

the maximum height of the inundated tsunami, water

density, and g acceleration due to gravity. Accordingly, maximum vertical sustained pressure distribution is
the same as hydrostatic pressure distribution. They also proposed the following pressure when the inundated
tsunami undergoes soliton fission, resulting in a bore:
p z g max max5.41 z 1.35 max ,31 z 3 max

for 0 z max 3

(2)

This apparently represents maximum sustained pressure, but because pressure is force per unit area,
pressure distribution is required.
Iizuka and Matsutomi (2000) proposed force associated with the steady state flow:
FH 1 2 C D u 2 h f B h

(3)

FH is the horizontal sustained force on the structure, u tsunami inundation velocity, C D the drag
coefficient (=1.1 to 2.0), h f tsunami inundation height in front of a structure, and B h the structures width.

u is proportional to the square root of tsunami inundation height, so FH is proportional to its square.

Part of Eq. 2 appears to be maximum impulsive pressure, but this is unclear. Impulsive pressure vertical
distribution and magnitude are unknown. Eq. 3 indicates that force is proportional to inundation speed, but
the effect of velocity is not included in Eq. (1) or (2).

3. Structural Performance Test of Walls In a Tsunami

3.1 Wooden Wall Tests
(1)

Large hydro-geo flume

The size of the large hydro-geo flume is a maximum of 184m long, 3.5m wide and 12m deep. The wave
paddle has a 14m stroke with piston type wave maker and generates a tsunami wave 2.5m high. The facility
generates either positive or negative tsunami waves. Fig. 3 shows the relationship between the wave paddle
stroke and positive tsunami height. Tsunami height is defined as the wave crest elevation above the still water
level. H T is tsunami height, S stroke, and T D wave paddle motion duration. Wave paddle motion is
based on the sine function as shown in Fig. 4. Fig. 5 shows the experimental arrangement cross-section for
model tests. The 1/10 slope was in the middle of the wave flume, followed by a flat part 20m long.
3
(m)

TD=10
TD=20
TD=30
TD=40

HT

0
6

10

S (m)

12

14

Figure 3: Relationship between stroke and tsunami height

3
- 53-

Positive tsunami
Negative Tsunami

Figure 4: Example of wave paddle motion (above) and tsunami profile (below)

Wave Paddle

Wave
Gauges

24.0

Slope 1/10

z
x

26.0

2.6

40.0

30.0

20.0

Unit
(m)
(m)
3.0

184.0

Figure 5: Cross-section of experiments

(1) Experimental setup
A wooden wall simulating a wooden house wall was placed in the flume to investigate the mechanism of
land-structure destruction due to tsunami force, as shown in Fig. 6. The 2.5m positive tsunami profile was
generated as shown in Fig. 4. The height of wall is 2.5m and width is 2.7m. The wooden wall is consist of
siding board, veneer board and wood column as shown in Figure 7. The flexural capacity of the siding board
with 16mm width and 45 cm span is 10 kN/m2 as short duration time.

Figure 6: Cross-section of experimental setup

Figure 7: Cross-section of the wooden wall

4
- 54-

(2) Destruction
Photo 2 shows the post-tsunami wooden wall And Photo 3 a plan view at the instant the tsunami front
reaches the wall. Note that the wall is destroyed at the moment water hits the wall.

Photo 2: The state of destruction of wooden wall

Photo 3: Plan view of tsunami hit instant

3.2 Concrete Wall Tests
(1) Cross-sections
Fig. 8 shows cross-sections of concrete walls 2.5m high and 2.7m wide, ranging in thickness from 6cm to
10.0cm. Wall specifications are shown in Table 1. Rebar 6 mm in diameter is located at 20 cm intervals both
vertically and horizontally. Both sides of a panel are fixed in 30 cm thick pillars. Pressure gauges are on walls
and strain gauges on surfaces and rebar. P denotes pressure and G strain gauges.

- 55-

G19 P6
P5
G13 G3 G15
P4
G17
G11
G5
P3
G7
P2

G20

G1

G21

G9

G2

G16 G4 G14
G18
G12
G6

G8
G10

P1

G22

Front view

Back view

D6@200
G1,2

G19,20

G3,4
A

G11,12 G5,6

G17,18

G13,14

G15,16

G21,22

A
A-A

Cross section (A-A)

G7,8
D6@200

G9,10
D6@200

Unit (mm)
(mm)

Bar arrangement

Figure 8: Cross-section of concrete wall test pieces

Table 1 Concrete wall specifications
Case No.

Wall thickness (mm)

Compressive strength (N/mm2)

60

21

60

33

75

21

75

33

80

32.7

90

32.7

100

21

100

32.7

(2) Pressure distribution

The water surface elevation time history is shown in Fig. 4 and vertical pressure distribution in Fig. 9.
Maximum impulsive bore pressure is 14 tons at the bottom which is about 7.7 times the hydrostatic pressure
associated with the inundation depth in front of the wall.

6
- 56-

2.0

Maximum
Impulsive

Bore Pressure

1.8

1.6

Maximum Sustainable
Pressure

z (m)

1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0

20

40

60

80 100 120 140 160

P (kN/m2)

Figure 9: Vertical pressure distribution

(3) Experimental results
The tsunami pressure time history for case 2 is shown in Fig. 10. Impulsive tsunami force at P1 to P3
was 43.9s. Destruction at time points A and B in Fig.4 is shown in Photo 4, taken from behind the wall using
a high-speed camera. Cracks progressed from bottom to top at time point A, with the wall broken when the
tsunami hit the lower wall.
Photo 5 shows the failure of the reinforced concrete wall with 6cm and 10cm wall thicknesses from the
front. Fig. 11 illustrates the difference in failure mode according to wall thicknesses. Under this tsunami
condition, if the thickness of wall is more than 10cm, the column will likely fail.

Fig. 10Tsunami pressure time histories (left: wide range, right: detailed)

7
- 57-

Photo 4 Destruction at time points A (left) and B (right)

Photo 5Concrete wall failure for 6cm (left) and 10cm (right) thicknesses

Fig 11 Differences in concrete wall destruction

8
- 58-

4. Concluding Remarks
Wooden house walls and concrete walls 6 cm and 10.0 cm thick in a flume to investigate the land-structure
destruction mechanism of tsunami inundation force broke the instant the 2.5m tsunami hit walls.
The difference of the failure mode of concrete wall due to impulsive tsunami force was clarified. If the
strength of wall is small, then the bending failure or punching shear failure is occurred. When the strength of
wall is increasing, the failure mode is shifting local failure to whole destruction.

References
Matsutomi, H. (1991), Distribution of impulsive pressure and impulsive force due to the collision of
bore, Proceedings of Coastal Engineering, JSCE, 38(1), 626-630. In Japanese.
Iizuka, H. and Matsutomi, H. (2000), Damage due to the Flooding Flow of Tsunami, Proceedings of
Coastal Engineering, JSCE, 47(1), 381-385. In Japanese.
Asakura, R., Iwase, K., Ikeya, T., Kaneto, T., Fujii, N. and Omori, M. (2000), An experiment study on
wave force acting on on-shore structures due to overflowing tsunamis, Proceedings of Coastal Engineering,
JSCE, 47(1), 911-915. In Japanese.

9
- 59-

- 60 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

NECESSITY OF ADVANCED TSUNAMI DAMAGE INDEX

Koji FUJIMA
Department of Civil and Environmental Engineering, National defense Academy, Yokosuka, Japan,
fujima@nda.ac.jp
Abstract
Hydraulic experiments were carried out for estimating tsunami wave force for onshore
rectangular structure. Building models were placed at several distances from a shoreline.
Wave pressure was measured at several points on the exposed structure. Because impact
and standing-wave pressures at each point reach the maximum values at different instants,
estimation of tsunami force by integrating an envelope of maximum wave-pressure
distribution tends to overestimate total force. Thus, measured force was used to formulate
an estimation equation for tsunami force. Hydrostatic formula was successful for cases of
structure near a shoreline, although it had a large scattering for cases of the structure far
from a shoreline. Hydrodynamic formula was successful for all cases, although an inertial
force was not negligible for cases of structure near a shoreline. These results showed the
necessity of advanced tsunami damage index.
Keywords: Tsunami force, design formula, damage estimation, wave pressure, drag
coefficient

1. Introduction
The 2004 Sumatra Earthquake and Indian Ocean Tsunami gave a vivid description of the menace of
major tsunamis. Most buildings in coastal areas were destroyed by the tsunami, and this wreckage became
the weapons which destroyed other buildings. Thus, it is important for tsunami-damage reduction that
buildings near a shoreline withstand tsunami. Besides, a solid structure near a shoreline can be utilized for
an evacuation building.
Many researches had been conducted to investigate tsunami force acting on structures, and many
estimation equations for tsunami force have been previously proposed. Most of them (e.g. Asakura et al.,
2000; Yeh, 2007; Simamora et al., 2007) introduced inundation depth value where there is no structure for
evaluation of external force. Yeh and Simamora et al. considered both inundation depth and velocity when
there is no structure. In addition, Simamora et al. examined using inundation depth at the front of a
building based on the assumption of hydrostatic pressure. In this paper, these are classified into two
groups. One is `equations in hydrostatic form' where only inundation depth is considered as the variable,
because an equation of this type has a similar expression to that based on the assumption of hydrostatic
- 61 -

pressure. Another is `equations in hydrodynamic form' where both inundation depth and velocity are
considered.
At present, inundation depth is widely used as an external force index for tsunami damage estimation
(Shuto, 1992). Therefore, if hydrostatic equations work well, it may support the validity of classical
tsunami damage estimation. However, we cannot sufficiently understand applicability and characteristics
of these formulas at present. In this paper, fundamental experiments were conducted to understand
characteristics of time history of wave pressure and total force. Besides, applicability of tsunami force
formulas was checked by the data of Yeom et al. (2007, 2008) in addition to the present data.

2. Experimental Setup
Hydraulic experiments were carried out in a two-dimensional wave basin with 11 m length, 7 m width,
and 1.5 m depth. This basin is equipped with a piston device for generating waves. The sea bed in the
basin was set to model a sea floor deformation from offshore, shallow water and to onshore areas. The
experimental setup is shown in Fig. 1.
Several cases of experiments were conducted by varying the scale of a building's model, distance from
shoreline, and stroke type of incident wave. There were two kinds of model used in the experiments. One
was the model of B = W = H = 10 cm where B = breadth in alongshore direction, W = width in
cross-shore direction and H = height. Another was of B = 20 cm, W = H = 10 cm. The distance from
shoreline to structure, D, was set to 20, 50, 80 and 150 cm. The wave paddle was programmed to move
backward slowly once and then move forward with stroke of the paddle set to 10, 15 and 20 cm. In all
cases, the incident wave had broken in shallow water area, and hit the vertical wall (seawall). Then, a part
of wave inundated the onshore area, and hit the vertical wall (structure model) again. The wave profile in
the offshore and shallow water areas and inundation depth in onshore area (with and without obstacle
model) was measured by wave gages. Wave velocity was measured by propeller current meter, wave
force on the model by load cell, and wave pressure by pressure gages. Inundation depth and velocity in
the onshore area were also measured with the same conditions of runup distance and wave stroke without
structure.
Characteristics of wave pressure on structures placed 20-cm onshore from the shoreline were
investigated. Only wave stroke of 15 cm was applied to avoid wave overtopping the model. Pressure
gages were mounted for several lines as shown in Fig. 2 for observing wave pressure for each element on
the exposed area of the model. Visual observation using high speed cameras (250 fps) was conducted to
examine the relation of the pressure and the wave form in front of the model.

3. Experimental Result Analysis and Discussions

3.1 Characteristics of time-history of wave pressure and total force
Fig. 3 shows time history of the wave pressure along line-1 and the total horizontal wave-force
measured by load cell. Every curve was obtained by ensemble average of five repeated measurements.
The wave force starts increasing earlier than the pressure because we could not set the pressure gage close
enough to the bottom.

- 62 -

45

257

55

Land

Sea
Wave paddle

695

2.5

60

1.4

B=10,20cm
W=10cm
H=10cm

1:3
0.8

181

175

383

3.75

2.2

80
3

2.5

585

B
W

2.5

370

P1

P1

P1

P2

P2

P2

P2

1.4

P3

P3

P3

P3

1.4

P4

P4

P4

P4

1.4
2.2

P5

P5

P5

P5

Line1

Line2

Line3

Line4

20.0

Unit:cm

Figure 1: Experimental setup.

1.8

P1

1.4

Unit:cm

Figure 2: Pressure gage set up on the model.

There appear two peaks in the time history of the wave pressure at P5. The visual observation by high
speed camera shows that the first peak at P5 occurred approximately when the splash in front of the
structure reached the maximum height and the second peak occurred approximately when the splash
rushed in with the main flow. We consider that the first peak is the impact pressure and the second peak is
the maximum standing-wave pressure. The maximum pressure at the upper position appears at 6.2 to 6.3 s,
almost the same time as the second peak at P5. However, there is a small time-gap. The time gap exists in
not only height-direction, but also breadth-direction. The time gap of wave pressure was caused by the
complicated motion of the wave hitting the structure and the irregularity of the wave. This fact is also
supported by the visual observation in Fig. 4; the wave form in front of the structure is clearly not
uniform in breadth-direction.

500

Standing wave
Pressure

Impact Pressure

3.0
P1
P2
P3
P4
P5
Load Cell

300

2.5
2.0
1.5

200

Force (N)

Pressure (Pa)

400

1.0
100

0.5
0

0.0
5.0

5.5

6.0

6.5

7.0

7.5

8.0

Time ( s )

Figure 3: Time history of wave pressure at line 1 for B = 20 cm model (center line).

- 63 -

(a)

(b)

Figure 4: Images of splash in front of the model ((a) B = 10 cm; (b) B = 20 cm).
Asakura et al. (2000) proposed the empirical equation to estimate maximum tsunami force by
integrating the envelope of maximum standing-wave pressures. Fig. 5 shows the maximum pressures
(including impact pressure) observed in the present measurements. The envelope of maximum pressures
obtained by Asakura et al. is also drawn in the figure where is the density of water, g the gravitational
acceleration, z the measurement height from the bottom, p the pressure, and hi the inundation depth at the
point of interest when there is no obstacle. It should be noted that the subscript 'm' expresses the
maximum value in the measured time history, so that pm denotes the maximum pressure and him the
maximum inundation depth at the point of interest when there is no obstacle. This figure shows that the
results of the present experiments were similar to those of Asakura et al. regarding the maximum wave
pressure distribution. However, as described before, measured pressures do not reach maximums in all
points at the same time. In addition, the total wave force reaches its maximum between the times of
maximum impact and standing-wave pressures. Thus, integration of the envelope of maximum
standing-wave pressures possibly overestimates tsunami forces, while it cannot guarantee that the
obtained equation does not underestimate the wave force.

z
him

pm
ghim
Figure 5: Non-dimensional relationship between the maximum wave pressure (pm) and the measurement
height (z).

- 64 -

3.2 Estimation of tsunami force in hydrostatic form

Several empirical formulas to estimate tsunami force have been proposed in hydrostatic form as
follows:
2
FS = ghim
B

(1)

where is a coefficient. Asakura et al. integrated the envelope of maximum standing-wave pressures, and
proposed = 4.5 for estimating maximum tsunami force. Tanimoto et al. (1984) proposed the maximum
standing-wave pressure distribution for the case of a breakwater. Integrating Tanimoto's distribution
above sea level yields = 3.3. Both Asakura's and Tanimoto's equations are aimed at establishing the
safety-estimation formula for structural design. Namely, it was important for them not to underestimate
tsunami forces.
The measured maximum tsunami forces are compared with the estimated force by Asakura's equation
in Fig. 6. The experimental data by Yeom et al. (2007, 2008), Simamora et al. (2007), and the present
study are also plotted in this figure. Fm denotes the maximum value in the measured time history of total
horizontal force. The figure shows that the hydrostatic formula has very large scattering. Asakura's
equation overestimates the tsunami force twice on average; however, there are some cases in which
Asakura's equation underestimates the force.
Locations where a large scattering occurs are examined in Fig. 7. The ordinate of the figure is ,
which was computed inversely from the experimental results. This figure shows that in most cases are
in range of 1 to 3, with an average value of 1.9. However the data is scattered from 0.7 to 7 for him /D <
0.05 (for cases of structures far from a shoreline). This might be caused by a variety of phenomena in the
onshore area. From this view, the hydrostatic form is not appropriate for the estimation equation of
tsunami force in him /D < 0.05. It may suggest that the tsunami damage estimation based on only
inundation depth is inaccurate for structures far from a shoreline. An advanced tsunami damage index is
required to estimate damage of structures far from a shoreline. In addition, Asakura's equation ( = 4.5)
does not underestimate the force for him /D > 0.05, although it does provide an overestimated value.
Because the average value of for him /D > 0.05 is 1.8, Asakura's equation overestimates tsunami force
2.5 times on average. Tanimoto's equation ( = 3.3) seems more acceptable for the safety estimation for

him /D > 0.05 than Asakura's equation.

3.3 Estimation of tsunami force in hydrodynamic form
Another formulation of tsunami force exerted on a building has been conducted as a hydrodynamic
force (drag force). Drag force is a function of inundation depth, velocity, and drag coefficient which
varies depending on the shape of the exposed area of the building. Yeh (2007) modeled the drag force by
the following equation:

FD =

CD
B(hi ui2 )m
2

(2)

- 65 -

100
6
5
4

4.5ghim2 [N/m]

3
2

10
6
5
4
3
2

Yeom(2008)
Yeom(2007)
Simamora(2007)
Present Study

1
6
5
67

4 5 67

10

4 5 67

100

Fm/B [N/m]

Figure 6: Comparison of measured wave force with estimated force by Asakura's equation.

Yeom(2008)
Yeom(2007)
Simamora(2007)
Present Study
Asakura eq.(=4.5)
Tanimoto eq.(=3.3)

7
6

Fm/gh im B

5
4
3
2
1

Average 1.90

0.00

0.05

0.10
him/D

0.15

Figure 7: Variation of by him /D.

Figure 8: Variation of drag coefficient CD by him /D.

- 66 -

0.20

Figure 9: Comparison of measured wave force with estimated force by Eq.(3).

where ui is the velocity at the point of interest when there is no obstacle and CD = 2.0 for rectangular
objects (FEMA-CCM, 2005). Since the maximum inundation depth, him, and the maximum velocity, uim,
may not occur at the same time, the above expression seems valid for estimation of the maximum
momentum flux. However, when we analyze the external force based on field survey results, we cannot
expect to obtain the time history of hiui2. Thus, it is convenient if we have a serviceable tsunami-force
formula based on himuim2. In this study, we analyzed both of the possibilities of himuim2 and (hiui2)m.
Figure 8 shows the drag coefficient which was computed inversely from the experimental results. The
drag coefficient has some scattering. It is roughly 2 for cases of a structure far from a shoreline (him /D <
0.05), and it seems to increase as him /D increases. We can propose an approximate equation for force
estimation if the drag coefficient is regarded as a function of him /D as follows:
FD1 =

h
CD1
B(hi ui2 )m , CD1 = 2.0 + 5.4 im
2
D

(3)

Applying Eq.(3) gave an effective approximation as shown in Fig. 9. Scattering of the data shown in
this figure is about half of that in Figure 6.
There is a possibility that the increase of drag coefficient near a shoreline is caused by an inertial force.
Thus, Morrison's formula has also been applied to estimate wave force.
u
C
FD 2 = D 2 Bhi ui2 + CM BWhi i , CD 2 = 2.0 , CM = 1.0
2
t m

(4)

where CM is mass coefficient and W is width of the model in cross-shore direction. Fig. 10 shows that CD2
= 2.0 and CM = 1.0 is appropriate on average and Fig. 11 shows that the agreement of Eq.(4) does not
depend on him /D. Thus, increase of CD1 in Eq.(3) might be caused by the inertial force.

- 67 -

Figure 10: Comparison of measured wave force with estimated force by Eq. (4).

Figure 11: Variation of Fm / FD2 by him /D.

Estimation of the drag coefficient based on the maximum inundation depth and the maximum velocity
shows relatively small scattering as shown in Fig. 12. The following estimation equation gives
satisfactory results as shown in Figure 13.
FD 3 =

CD 3
h
2
Bhim uim
, CD 3 = 1.3 + 6.3 im
2
D

(5)

Figure 12: Variation of apparent drag coefficient by him /D.

- 68 -

Figure 13: Comparison of measured wave force with estimated force by Eq.(5).
The above examinations indicated that the tsunami damage estimation based on only inundation depth
was inaccurate for structures far from a shoreline. The estimation equations in hydrodynamic form in
Eqs.(3), (4) and (5) are successful for all cases. Thus, these parameters may be serviceable for new
tsunami damage index.
Equation (4) seems superior to the other two equations because the coefficients do not depend on
him/D. However, it is doubtful whether the time history of velocity is predicted accurately by the usual
tsunami numerical simulation. Applicability of the proposed equation should be examined including the
accuracy of prediction of inundation depth and velocity in the numerical simulation.
The upper limitations of data in Figs. 9, 10 and 13 are about twice the average values (Eqs.(3), (4) and
(5)) and the lower limitations are about half of the averages. The magnitude of these scatterings is almost
similar to that for him /D > 0.05 in Figure 7. Thus, one natural idea is that the hydrostatic formula is used
for him /D > 0.05 and the hydrodynamic form is applied for him /D < 0.05. If Eq.(3) or (5) is used for him /D
< 0.05, the increase of drag coefficient by him /D is not so important. Then, CD1 and CD3 may approximate
2.1 and 1.5, respectively, in this range. For example, it is possible to formulate the average estimation and
safety estimation of tsunami force as follows:

1.5
2.1
2
B(hi ui2 ) m or
Bhim uim

2
2
=
F
average estimation: m
2

1.8 ghim
B

2.6
4.7
2
B(hi ui2 ) m or
Bhim uim

2
safety estimation: Fm = 2
2

3.3 ghim
B

(him /D<0.05)

(6)

(him /D>0.05)
(him /D<0.05)

(7)

(him /D>0.05)

Note that the safety-estimation equations are obtained by formulating the upper limitation of data. The
coefficient in the safety-estimation equation for him /D < 0.05 was determined by the fact that the
maximum values of the drag coefficient in Figure 8 and 12 were about 4.7 and 2.6, respectively; and
Tanimoto's equation was adopted for safety estimation for him /D > 0.05.
Based on Eqs. (6) and (7), him is serviceable as tsunami damage index for him /D > 0.05. However,
- 69 -

(hiui2)m or himuim2 should be used for tsunami damage index for him /D < 0.05.

4. Concluding Remarks
The conclusions from this study can be summarized as follows:
1. The standing-wave pressure at different points becomes maximum at different moments in time
because of the complicated motion of the wave hitting the structure and the irregularity of the wave.
Moreover, the total wave force also becomes maximum at different moments. Thus, integrating the
envelope of maximum wave pressure possibly causes the overestimation of maximum tsunami
forces.
2. In hydrostatic form, Tanimoto's equation is more acceptable for safety estimation in him /D > 0.05 than
Asakura's equation. The hydrostatic form seems inappropriate to model tsunami force for him /D <
0.05. This fact suggests that the tsunami damage estimation based on only inundation depth is
inaccurate for structures far from a shoreline.
3. Hydrodynamic form was tried to reproduce the tsunami force. Three types of estimation equations
were proposed. Because an inertial force is not negligible for cases of a structure near the shoreline,
CD1 and CD3 are formulated as functions of him /D. As a result, these gave satisfactory results for all
cases.
4. New tsunami-force estimation formulas are proposed as in Eqs.(6) and (7). The applicability of
equations should be examined including the reproducibility of inundation depth and velocity by
numerical simulation.

References
Asakura, R., K. Iwase, T. Ikeya, M. Takao, T. Kaneto, N. Fujii and M. Omori (2000): An Experimental
Study on Wave Force Acting on On-shore Structures Due to Overflowing Tsunamis, Proc. of Coastal
Engineering, JSCE, Vol. 47, pp. 911-915, (in Japanese).
FEMA-CCM (2005): Coastal Construction Manual, FEMA 55 Report, Ed.3.
Shuto, N. (1992): Tsunami Intensity and Damage, Tsunami Engineering Technical Report, DCRC,
Tohoku Univ., No.9, pp.101-136 (in Japanese).
Simamora, C., Y. Shigihara and K. Fujima (2007): Experimental Study on Tsunami Forces Acting on
Structures, Annual Journal of Coastal Engineering, JSCE, Vol. 54, pp. 831-835 (in Japanese).
Tanimoto, K, H. Tsuruya and S. Nakano (1984): Examinations on Tsunami Force and Damage of
Bulkhead by 1983 Nihonkai-Chubu Earthquake Tsunami, Proc. of Coastal Engineering, JSCE, pp.
257-261 (in Japanese).
Yeh, H.(2007): Design Tsunami Forces for Onshore Structures, Journal of Disaster Research, 2(6), pp.
531-536 .
Yeom, G.-S., N. Mizutani, K. Shiraishi, A. Usami, S. Miyajima and T. Tomita (2007): Study on Behavior
of Drifting Containers Due to Tsunami and Collision Forces, Annual Journal of Coastal Engineering,
JSCE, Vol. 54, pp. 856-860 (in Japanese).
Yeom, G.-S., T. Nakamura, A. Usami and N. Mizutani (2008): Study on Estimation of Collision Force of a
Drifting Container Using Fluid-structure Interaction Analysis, Annual Journal of Coastal
Engineering, JSCE, Vol. 55, pp. 281-285 (in Japanese).
- 70 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

DETAILED AND REAL-TIME ESTIMATION METHODS

OF TSUNAMI

Takashi TOMITA
Tsunami Research Center, Port and Airport Res. Inst., Yokosuka, Japan, tomita@pari.go.jp

Daisuke TATSUMI
Tsunami Research Center, Port and Airport Res. Inst., Yokosuka, Japan, tatsumi@pari.go.jp

Kazuhiko HONDA
Nagoya Research and Engineering Office for Port and Airport, Chubu Regional Bureau, Ministry of Land,
Infrastructure, Transport and Tourism, Nagoya, Japan, honda-k852a@pa.cbr.mlit.jp
Abstract
Tsunami disaster mitigation would be initiated by imaging and recognizing possible
damages. Two methods are presented to estimate tsunami damages in this paper. One is the
numerical simulation system to estimate the detailed tsunami features which lead to realize
possible tsunami damages. In order to calculate the tsunami propagating in the open ocean,
inundating coastal areas and interacting with structures, a multi-scale and multi-models
system named STOC (Storm surge and Tsunami simulator in Oceans and Coastal areas) has
been developed. Validations of the models have been conducted through comparison with
hydraulic experimental results. The other method is the real-time tsunami estimation in the
coastal areas. To predict the tsunami profiles along coasts precisely and quickly, the
inversion method with the offshore tsunami observation data is improved by using the
location of the epicenter as a priori information. From numerical experiments conducted
around the Nankai Trough, Japan, the initial water surface elevation offshore and the
tsunami profiles along coasts can be predicted with good accuracy using the tsunami
profiles observed at 5 offshore locations for 15 minutes.
Keywords: Tsunami; numerical simulation; damage estimation; real-time estimation;
three-dimensional model; dynamic hazard mapping

1. Introduction
As shown by disasters due to the 2004 Indian Ocean Tsunami and recent other tsunamis, a tsunami
causes various types of damage to coastal areas such as inundation, destruction of houses, drift of vessels,
beach erosion and etc. To develop more effective plans and measures for disaster management including
- 71 -

business continuity after the disaster, we should predict and understand tsunami disasters as realistic as
possible. Furthermore, from storm surge disasters by the Hurricane Katrina in 2005, we have realized that
storm surge and waves strike weak points of coastal defense system and cause disasters even in a
protected city. Since we can easily image that the tsunami would cause disasters in a similar way, we
should also understand defense performance of existing and planning structures against tsunamis.
To predict and understand the tsunami propagating in the ocean we already have useful numerical
models, which are based on the long wave theory. In shallow water the wave nonlinearity is usually
considered. The wave dispersion should be significantly considered for the tsunami propagating a long
distance (Fujima et al., 2007). Application of a Boussinesq-type model to the tsunami in s coastal area has
shown that the wave dispersion is also important even in the coastal tsunami (Iwase et al., 2002).
Recently hydrodynamics of tsunami is also focused on to understand tsunami flow velocity, tsunami wave
load, interaction with structures and structural response in order to develop effective tsunami mitigation
measures. The Reynolds-Averaged Navier-Stokes equation (RANS) models, Smooth Particle
Hydrodynamics (SPH) model and Moving Particle Semi-implicit (MPS) model have been developed and
applied to estimate tsunami hydrodynamics. For example, the RANS model was applied to the tsunamis
running up on a hill surface with complicated configuration at the Monai area of the Okushiri Island
struck by the 1993 Okushiri Tsunami (Yoneyama et al., 2002).

Fujima et al. (2002) and Tomita et al.

(2006) applied their RANS models to calculate the tsunamis overflowing submerged sections of
breakwater, and obtained good results in comparison with physical model tests. Furthermore, Honda and
Tomita (2009) showed that non-hydrostatic effects appear in the tsunami obliquely overflowing a quay,
based on numerical simulations.
Precise and detailed numerical simulations are useful for understanding tsunami damages and provide
development of tsunami disaster mitigation measures with consideration of local characteristics such as
bathymetric and topographic features, arrangement of structures and buildings and social characteristics.
On the other hand, real-time simulation is also useful for secure evacuation of people and
disaster-reduction activities before tsunami arrival. The

tsunami

warning

system

by

the

Japan

Meteorological Agency (JMA) is very useful and effective to understand the tsunami striking coasts soon.
If we can estimate and understand inundation areas before tsunami arrival, it may be further useful for
disaster-reduction activities. Recently real-time estimation of tsunami have been investigated (Koike,
2002; Yasuda eta al., 2007; Tatsumi and Tomita, 2007), using the observation data of offshore tsunami
(Nagai et al, 2006)
In this paper are introduced two tsunami simulation methods to estimate precise and detailed tsunami
damages and to calculate the tsunami in coastal areas within a very short time after the earthquake
occurrence. The later model is part of a real-time tsunami inundation estimation model developed at the
present.

2. Numerical Modeling for Estimation of Tsunami Damages

2.1 Outline of STOC
Considering current computer performance, coupling of adequate numerical models is needed to
estimate and understand damages due to the expected tsunami, because the tsunami comes from an open
ocean and causes damage locally. Therefore, we have developed the STOC (Storm Surge and Tsunami
- 72 -

Simulator in Oceans and Coastal Areas) system (Tomita et al., 2006), which consists of three fluid models
named STOC-IC, STOC-ML and STOC-VF (CADMAS-SURF/3D) and a floating-bodies model named
STOC-DM. STOC-IC is a RANS model to calculate the detailed tsunami in a coastal area with
consideration of interaction between the tsunami and structures. STOC-ML is a hydrostatic model to
calculate the tsunami propagating in wide sea, and is combined with STOC-IC, as shown in Fig. 1.
STOC-IC is also connected to CADMAS-SURF to estimate impulsive wave pressure of the tsunami
acting on a structure. CADMAS-SURF is another RANS model with the VOF method to determine the
water free surface of tsunami. Arikawa and Yamano (2008) have shown that tsunami pressure including
the impulsive one is well calculated by CADMAS-SURF with a special treatment to remove numerical
noise in pressure calculation. STOC-DM is a practical model to estimate numerous tsunami-drifted bodies
such as vessels, cars and containers in actual ports and harbors (Honda et al., 2009).
Estimatetsunami
damagetoacoastal
city/town

STOCIC
Fully3D

STOCML
MultiLevelModel
hydrostaticpressure
ineachlevel
STOC-DM
Drifting Body
Model
Tsunami Source

STOCVF

Fully3D+VOF
Estimatedestructionof
structure

Estimatethetsunamis
propagatinginanocean
includinganinitialtsunamiprofile

Figure 1: STOC model.

2.2 STOC-IC
STOC-IC is a three-dimensional Reynolds Averaged Navier-Stokes (RANS) model which is applied to
estimate the tsunami in the coastal area more than tens of square kilometers wide, in order to estimate
three-dimensional flows of tsunami around structures. The governing equations are as follows:

( )

( )

( )

xu +
yv +
zw = 0
x
y
z

(1)

1 p
u

x uu +
y vu + z wu f 0 v = v
+
x
t x
y
z

u v

u
u w
+ x e 2 + y e + + z e +

x
x y
z x
y x z
v

(2)

1 p
+
x uv +
y vv +
z wv + f 0 u = v
t x
y
z
y

v u
v w

v
+ y e 2 +

+
+
+
x e
z e
x
y z
x y y
z y

- 73 -

(3)

v
+

1 p
+
g
x uw +
y vw +
z ww = v
t x
y
z
z
w v

w
w u
+
+ + z e 2
+ y e

x e
x

z
y
y
z
z

(4)

in which (x, y, z) are the Cartesian coordinates, (u, v, w) the water particle velocity components in the
directions of x, y and z, the fluid density, p the pressure, g the gravitational acceleration, e the eddy
viscosity, and fo the Coriolis coefficient. The porosity of v and transmissivity of x, y and z in the
directions of x, y and z introduced by Sakakiyama and Kajima (1992) are imposed to express
configuration of sea bottom and shape of structures smoothly. The eddy viscosity coefficient is estimated
by the following formula in the same way as Nakatsuji et al. (1992) and Fujima et al. (2002):

x j

u
e = (C s )2 i +

u j u i u j

+
xi x j xi

(5)

under the summation rule. In Eq. 5, = (xyz)1/3 and (x, y, z) the computational grid spaces in x,
y, and z directions, respectively. Cs is a constant vale equal to 0.2 in this study. Constant eddy viscosity
coefficient is also available for e. To reduce computational efforts to calculate the detailed tsunami in the
wide area, the free water surface is detected by means of the vertically integrated continuity equation in
the same way as Fujima et al. (2002).
The governing equations are solved by the finite difference method. A staggered mesh in space and
leapfrog method in time are applied. Conservation terms in the governing equations are discretized by
means of the control volume method. Diffusion terms are discretized with the second-order central
scheme, while the advection terms are expanded with a hybrid scheme in which the first-order upwind
scheme is combined with the second-order central scheme for stability. The pressure is calculated by the
method of Simplified Marker And Cell (SMAC).
2.3 STOC-ML

The governing equations of STOC-ML are almost same as those of STOC-IC. However, STOC-ML is
used the hydrostatic pressure assumption in order to reduce calculation efforts for the tsunami
propagating in open ocean.
2.4 Connection of STOC-IC and STOC-ML

STOC-IC and STOC-ML are connected each other in overlapping zones, in which the physical
quantities such as the water surface displacement, fluid velocity, wave pressure in each computational
area of STOC-IC or STOC-ML are adjusted, using the interpolation technique.
2.5 STOC-DM

STOC-DM is a practical model to calculate motion of numerous floating bodies by the tsunami. The
motion of each body is excited by drag and inertia forces depending on the fluid velocity calculated with
STOC-IC and STOC-ML, taking account of collision with other floating bodies and structures. Each body
is modeled by a rigid rectangular for easy computation. The drag and inertia coefficients to estimate
tsunami wave forces acting on the rectangular are based on the experimental values introduced by Ikeya
- 74 -

et al. (2005).
2.6 Validation of STOC

(1) Tsunami passing breakwaters opening section with a submerged breakwater

The tsunami breakwater has usually a submerged breakwater in its opening section in order to reduce
the mass flux of tsunami intruding into the area behind it. Tanimoto et al. (1988) carried out physical
model tests in an indoor wave basin to investigate the tsunami reduction effects of such a breakwater. His
experimental basin is reproduced by means of STOC-IC and STOC-ML. The experimental basin and
cross section of the submerged breakwater in the opening section are shown in Fig. 2. This wave basin is
reproduced by 0.04 m grids around the submerged breakwater which is covered by the dotted line in Fig.
2, and 0.12 m grids outer the area. In the computation area where STOC-IC is applied, the vertical grid
size is 0.12 m from the bottom to the top of the submerged breakwater mound, and 0.04 m above the crest
of the mound. In the calculation area by STOC-ML the water area is not divided vertically and is set a
single layer from the bottom to the water surface. It should be noted that the value of 0.04 m corresponds
to 2 m in actual scale of the field.
Since the tsunamis were produced as uniform flows in Tanimotos experiments, the averaged velocity
at the opening section is set to 1.01 m/s in the computation, corresponding to the experimental cases of
0.98 m/s and 1.05 m/s.

Figure 2: Outline of experimental basin (left) and submerged breakwater in the A-A section (right).
Before comparing the calculation results with the experimental results, the effects of the coefficient of
Cs in the eddy viscosity model of Eq. 5 and constant horizontal eddy viscosity coefficient of eh in
STOC-ML are investigated through a parameter study, as shown in Fig. 3. Even if the value of Cs is
changed from 0.13 to 0.25, discrepancy among the calculation results is little. The calculation results are,
therefore, not sensitive to the change of Cs in value. In the same way, there are not much differences
among the results by STOC-ML in the range of eh from 10-6 to 10-3 m2/s. However, it should be noted
that the water surface profile of STOC-ML is much different from that of STOC-IC. Based on these
comparisons, therefore, the default values of Cs and eh are set as 0.2 and 10-3 m2/s, respectively.

- 75 -

Figure 3: Water surface profile deepening on Cs (STOC-IC) and eh (STOC-ML).

Fig. 4 shows the comparison between the calculation results of STOC-IC with Cs = 0.2 and STOC-ML
with eh = 10-3 m2/s and the experimental results. The broken and dotted lines indicate the experimental
result with the averaged velocities of 1.05m/s and 0.98m/s, respectively. STOC-IC is able to calculate the
water surface profile around the submerged breakwater more accurately than STOC-ML. Especially
STOC-IC can represent the wavy water surface profile behind the breakwater. It should be noted that the
wavy profile was not able to reproduce even by STOC-IC in the previous calculation with coarse
computation grids of 0.25 m around the submerged breakwater. The grid size of 0.04 m in this calculation
is 2 m in an actual field in rough estimation.

Figure 4: Comparison on water surface profile around the submerged breakwater.

(2) Flow around a rectangular weir
STOC-IC and STOC-ML are further applied to calculation of flows around a rectangular, which were
investigated experimentally by Hom-ma (1940). A channel in the calculation is illustrated in Fig. 5. The
value of q (m2/s) is the inflow discharge rate per unit width, and h1 and h2 (m) are the water surface
- 76 -

elevation above the top of weir at the upstream and at the downstream, respectively. The model scale of
the calculations is the same as the physical experiments. The discharge rates of 0.01, 0.02, 0.03 and 0.04
m2/s are set based on the experimental conditions. The horizontal grid size of whole computational
domain is gradually changed from 0.01 m around the weir to 1 m at each end of the channel. The channel
is vertically divided by grid size of 0.01 m for the computation by STOC-IC, whereas a single layer is set
from the bottom to the surface for the case of STOC-ML. The time step of marching calculation is 0.002 s,
depending on the CFL condition.

Figure 5: Numerical channel with a rectangular weir.

Hom-ma (1940) proposed the following well-known formula of discharge coefficient for the
rectangular weir based on his experimental results.
(6)

C = q h1 2gh1

0.35h1 2 gh1
q=
0.91h2 2 g (h1 h2 )

(h2 / h1 2 3)
(h2 / h1 2 / 3)

(7)

in which C is the discharge coefficient. The experimental and numerical results are compared in the
discharge coefficient, as shown in Fig. 6. The results of STOC-IC are good agreement with the
experimental results and Hom-mas formula, whereas STOC-ML provides slightly bigger values in the
cases of h2/h1 > 0.6 than the formula. Furthermore, the water surface profile above a wider weir 1 m wide
is shown in Fig. 7 to make the difference of STOC-IC from STOC-ML clearly. In the case of q=0.03 m2/s
shown as the figure, the critical water depth above the weir is 0.045 m theoretically, which is well
calculated by STOC-IC.

Figure 6: Comparison on discharge coefficient.

- 77 -

Figure 7: Comparison on critical depth of a broad-crested weir.

2.7 Application of STOC System to Actual Bathymetry and Topography

The STOC system is applied to actual bathymetry and topography. The topographic data is
constructed using the airborne laser scanning survey data with spatial resolution of 1 m approximately. In
the calculation, houses and structures are considered as rigid bodies, which reflect and diffract the
tsunami, since existence of houses in the tsunami inundation simulation is effective for reproduction of
the actual inundation area in the numerical simulation (Tomita and Honda, 2009). In the calculation, the
nesting grid system is used to calculate tsunami propagation in the open sea and inundation in a port area
simultaneously. STOC-IC is applied in Area-7, as shown in the right figure of Fig. 8, with the finest
horizontal grid of 2 m and nine layers of 1 m to 2.5 m high in the computation. In the other calculation,
STOC-ML with a single layer is applied in the same Area-7. The coarsest horizontal grid of 1458 m is set
in the open sea including the tsunami source as Area-1, and the grid size is gradually reduced to Area-7
with the nesting ratio of 1/3. STOC-ML is applied in Area-1 to 6 for both calculations.

Figure 8: Topography of Area-6 and Area-7.

Since the first tsunami is estimated to be largest model area in the model area, it is calculated. The
tsunami height is approximately 3 m. Fig. 9 shows the distribution of maximum inundation depth
overland and difference in the maximum inundation depth between STOC-IC and STOC-ML.
Discrepancy between them is not so large in the maximum inundation depth. However, difference
between them appears in the fluid velocity, as shown in Fig. 10 which indicates the difference between
the maximum horizontal velocities of STOC-IC and STOC-ML. There are especially larger differences
especially appear around the edge of channel and bending part of channel. The faster fluid velocity by the
non-hydrostatic model than the hydrostatic model has appeared in a meandering compound channel

- 78 -

(Fukuoka et al., 1998). The fluid velocity has important influence in occurrence and motion of floating
bodies by the tsunami.

ICML)

Figure 9: Maximum inundation depth (left) and difference between the calculated maximum inundation
depth by STOC-IC and STOC-ML (center).

IC ML)

Figure 10: Difference between the calculated maximum velocity by STOC-IC and STOC-ML.

Figure 11: Vessel motion excited by the tsunami 3.5 minutes (left), 5.0 minutes (center) and 6.5 minutes
(right) after the start of drift motion. The arrows indicate fluid velocity vectors around the initial water
level.

- 79 -

Fig. 11 shows motion of the vessels moored in the port calculated by STOC-DM. The vessel whose
mooring system is broken by the tsunami load acting on each vessel and the water surface elevation by
the tsunami starts to move and its motion is excited by the drag and inertia forces depending on fluid
velocity.

3. Real-Time Estimation of Coastal Tsunami with the Inversion Method

3.1 Outline of real-time tsunami estimation with the inversion method

(1) Principle of the estimation method

At the points with water depth over 50 m, the momentum equation of tsunami propagation is
approximated to a linear equation, so tsunami profiles are assumed to be the superposition of unit
tsunamis which are the tsunamis generated by 1 m uplift of the small segments of the divided tsunami
source area. The inversion method is based on this superposition principle and consists of the following 3
steps:
1. Divide the assumed tsunami source area into small segments. Calculate the propagation of unit
tsunamis and store the data of the wave profiles of each unit tsunami at certain locations in a database.
This first procedure needs to be finished before the occurrence of earthquakes.
2. When an earthquake occurs and the tsunami is observed at offshore locations, estimate the initial
water surface elevation (the tsunami source) by fitting the summation of unit tsunamis to the measured
tsunami profiles at offshore locations with the least squares method.
3. Predict the tsunami profiles along coasts by the superposition of unit tsunamis according to the
estimated initial water surface elevation.
The square residual between the summation of unit tsunamis and the measured tsunami profiles is
described as E in Eq. 8:
E = Ha

(8)

~1,1 L ~1, N
1
a1

= M , a = M , H = M O M
~

~

a
M
N
M ,1 L M , N

(9)

in which j is the measured water surface elevation at a certain point A at the time step j (j = 1 ~ M), ai is
the initial displacement of the segment i (i = 1 ~ N), and ~j,i is the water surface elevation at the point A
at the time step j generated by 1 m uplift of the segment i. The initial displacement of segments (the
vector a) to minimize the square residual is given by the solution of the following simultaneous linear
equations known as normal equations:
(10)

H T Ha = H T

The advantage of the inversion method is short computation time because all have to be done is just to
solve the simultaneous linear equations (Eq. 9) and the superposition of unit tsunamis. On the other hand,
the inversion method cannot be applied to the propagation in shallow water and runup because
nonlinearity effect cannot be ignored in shallow water and on the land.
(2) Improvement of the inversion method by the location of the epicenter
The simultaneous linear equations (Eq. 9) are often ill-posed systems, so Koike (2002) utilizes a priori
information that the initial water surface elevation is smooth in spatial direction. Smoothness of the initial

- 80 -

water surface elevation is represented by the spatial differentiation of the initial displacement of segments,
and this spatial differentiation needs to be minimized along with the square residual.
To shorten the length of observation time and improve the accuracy of prediction, the present study
utilizes other a priori information that the tsunami source is located near the epicenter. The location of the
epicenter can be determined in several minutes after the occurrence of an earthquake by the analysis of
seismic waves, so the utilization of the location of the epicenter is practical method.
The initial displacement of segments (the vector a) to minimize the distance between the epicenter and
the tsunami source together with the square residual and the spatial differentiation of the initial water
surface elevation is given by the solution of the following equation:

+ 12

Da

+ 22

Fa

)=0

L
4 1

0.0

4 1 L

D=
, F =
0.1

0.2

4
1

(11)

in which the term Da represents the spatial differentiation of the initial displacement of segments, and the
term Fa represents the distance between the epicenter and the tsunami source. Matrix F is a diagonal
matrix, and the diagonal elements correspond to the distance from the epicenter. 1 and 2 are the
weighting coefficients for a priori information. The sum of the weighting coefficients for a priori
information is assumed to be equal to the weighting coefficient for the observed data (the square residual),
so 1 and 2 are set to 0.5 respectively.
The following simultaneous linear equations are derived from Eq. 11, and the solution of Eq. 12 gives
the initial displacement of segments.

(H

(12)

H + 12 DT D + 22 F T F a = H T

3.2 Numerical Experiments around the Nankai Trough, Japan

(1) Outline of numerical experiments

To validate the improved inversion method, numerical experiments are conducted in Japanese coastal
area along the Pacific (Fig. 12). The area includes the subduction zone named the Nankai Trough, where
large earthquakes often occurred. There is not enough offshore tsunami observation data, so the observed
tsunami profiles are substituted by the numerical calculation based on the fault parameters of 4 historical
earthquakes (Table 1, Sato 1989 and Koshimura 2004). The numerical model is the two-dimensional
linear long wave model. Spatial resolution is 1350 m, and time step is 1 s. The size of the calculation
domain is 800 km by 440 km. Tsunami propagation is calculated until 3 hours after the occurrence of
earthquakes.
Tsunami profiles are assumed to be observed at 5 offshore locations. Based on the specification of GPS
Buoy, these observation points are located at most 20 km off the coasts. Although GPS Buoy can actually
measure the water surface elevation at every 1 s, the water surface elevation at every 10 s is used in the
present study.

- 81 -

34
23
19
18

21 22
20

16
3

15

25

13
8

11
9

32

28

35

38
37
36

33

26

39
40
41

29 31

10

17

27

12

14

24

30

A
2
1

Figure 12: Calculation domain. The area surrounded by a dashed line is the assumed tsunami source area.
Filled circles (A~E) indicate GPS Buoy, and white triangles (1~41) indicate the sample points for the
prediction respectively.
Table 1: Four historical earthquakes.
Earthquake

Year

Magnitude

Location of the epicenter

Hoei

1707

8.4

33.2 N, 135.9 E

Showa Tonankai

1944

7.9

33.80 N, 136.62 E

Showa Nankai

1946

8.0

33.03 N, 135.62 E

Tokaido-Oki

2004

7.4

33.2 N, 137.1 E

(2) Arrangement of the segments and unit tsunamis

The assumed tsunami source area is divided into 811 segments of 13500 m square (Fig. 12), based on
Kawata and Koike (1997), in which the size of a segment is set to be 10 times as large as the spatial
resolution. To generate unit tsunamis, 1 m uplift of a segment is given in the curved shape. The unit
tsunamis generated by the curved shape uplift can exclude high frequency noise, compared with those by
the column shape uplift (Koike 2002). In addition, the curved shape uplift is confirmed to be suitable for
reproducing the tsunami source with complex shape.
(3) Validation of the improved inversion method
The tsunami profiles observed at 5 offshore locations for 15 minutes after the occurrence of
earthquakes are inputted to the inversion method. Table 2 shows the comparison between the inversion
method with and without earthquake information. The accuracy of prediction is evaluated by the root
mean square error (RMSE) of the initial water surface elevation. Based on the superposition principle, the
tsunami profiles along coasts can be predicted with good accuracy if RMSE is small enough. The
numbers in parentheses following RMSE indicate the ratio of RMSE to the maximum value of the true
initial water surface elevation. Condition number and Dirichlet spread function will be explained later.
The improved inversion method can estimate the initial water surface elevation with RMSE of around
10 % of the maximum value of the true initial water surface elevation. RMSE is reduced 10 ~ 70 % by
utilizing the location of the epicenter. The improved inversion method is confirmed to be effective
especially for the vast tsunami source areas such as the Hoei Earthquake and the Showa Nanakai
Earthquake.
- 82 -

Table 2: Comparison between the inversion method with and without earthquake information.
Earthquake

Evaluation indicator

With earthquake info.

Without earthquake
info.

RMSE (m)

0.53 (8 %)

8.60 (129 %)

Condition number

69

1759

Dirichlet spread f.

804

2775

Showa

RMSE (m)

0.12 (12 %)

0.17 (18 %)

Tonankai

Condition number

82

1759

In 1944

Dirichlet spread f.

810

2775

Showa

RMSE (m)

0.27 (11 %)

1.89 (78 %)

Nankai

Condition number

70

1759

In 1946

Dirichlet spread f.

803

2775

RMSE (m)

0.06 (8 %)

0.08 (10 %)

Condition number

98

1759

Dirichlet spread f.

819

2775

Hoei
In 1707

Tokaido-Oki
In 2004

As an example, the results for the Showa Tonankai Earthquake are explained in detail. Fig. 13 shows
the estimation results for the initial water surface elevation. Although the shape of the tsunami source is
not necessarily estimated with good accuracy, the location and the maximum water surface elevation of
the tsunami source is estimated precisely. The tsunami arrival time and the first tsunami height at 41
sample points along coasts are shown in Fig. 13. The tsunami arrival time is defined as the time when the
water surface elevation change exceeds 1 cm. The first tsunami is defined as the tsunami whose tsunami
height is over 5 cm. The tsunami arrival time is predicted precisely at almost all the sample points. The
first tsunami height is also predicted with good accuracy except for at the sample points No. 2 ~ 4 and No.
34 ~ 40.
Fig. 15 shows the prediction results for the tsunami profiles along coasts. As mentioned in Fig. 14, the
tsunami arrival time and the first tsunami height are predicted with good accuracy especially at the
sample points No. 9, No. 29, and No. 32. The length of observation time is 15 minutes after the
occurrence of the earthquake, and the calculation time for the prediction is only a couple of minutes. Thus,
the tsunami prediction information can be disseminated to the residents before the peak of the first
tsunami arrives.

-0.5

0.0

0.5

1.0 (m)

0.2

0.2
0.2

Figure 13: The true (left) and the estimated (right) initial water surface elevation for the Showa Tonankai
Earthquake. Contour lines are plotted with the interval of 0.2 m.

- 83 -

Water Surface Elevation (m)

Figure 14: The first tsunami height and the tsunami arrival time for the observed and the predicted
tsunami profiles along coasts for the Showa Tonankai Earthquake.
0.4

0.2

0.5

0
-0.5

-0.2

No. 9

-0.4

No. 29

-1

0.5

-2

-0.5

No. 32

-4
0.5

No. 34

-1
1
0.5

0
-0.5

-0.5

No. 38
0

30

60

90

120

150

180

-1

GPS Buoy D
0

30

60

90

120

150

180

Time (minute) after the Occurrence of Showa Tonankai Earthquake

Figure 15: The observed (solid line) and the predicted (dot line) tsunami profiles at 5 sample points and
GPS Buoy. Their locations are shown in Fig. 12.
(4) Parametric studies
To investigate how the parameters such as the length of observation time and the number of offshore
observation points affect the accuracy of prediction, parametric studies are conducted.
Length of Observation Time
The length of observation time is varied from 5 minutes to 60 minutes after the occurrence of
earthquakes. The ratio of RMSE to the maximum value of the true initial water surface elevation is
plotted against the length of observation time in Fig. 16. Long observation time is confirmed to reduce
RMSE and improve the accuracy of prediction. However, RMSE for the Hoei Earthquake seems not to be
reduced by long observation time. It is because the observed data is dominant for the solution of the
inversion method when observation time becomes long. A priori information cannot be utilized
sufficiently because the weighting coefficients for a priori information (1 and 2 in Eq. 12) are fixed at
0.5 and do not change according to the length of observation time. As mentioned in Table 2, the
- 84 -

utilization of a priori information is important especially for estimating the vast tsunami source areas like

Ratio of RMSE to the Maximum Value of

the True Initial Water Surface Elevation

the Hoei Earthquake.

Hoei

Tonankai

Nankai

Tokaido-Oki

0.15
0.13
0.11
0.09
0.07
0.05
0.03
0

10

20

30

40

50

60

Length of Observation Time (minute)

Figure 16: Ratio of RMSE to the maximum value of the true initial water surface elevation versus the
length of observation time.
Number of Offshore Observation Points
The number of offshore observation points is varied from 3 to 19. Offshore observation points are
assumed to be installed with even intervals along the coast, so the interval is set to be 25 km for 19, 50
km for 10, 100 km for 5, and 200 km for 3 offshore observation points respectively. The ratio of RMSE to
the maximum value of the true initial water surface elevation is plotted against the number of offshore
observation points in Fig. 17. A large number of offshore observation points are confirmed to reduce

Ratio of RMSE to the Maximum Value of

the True Initial Water Surface Elevation

RMSE and improve the accuracy of prediction.

Hoei

Tonankai

Nankai

Tokaido-Oki

0.15
0.13
0.11
0.09
0.07
0.05
0.03
0

10

15

20

Number of Offshore Observation Points

Figure 17: Ratio of RMSE to the maximum value of the true initial water surface elevation versus the
number of offshore observation points.

4. Conclusions
Two estimation methods have been developed to create and enhance tsunami damage images, which
initiate disaster mitigation plan and measures of possible tsunamis.

- 85 -

One is the numerical models system named STOC to estimate the detailed damage by the possible
tsunami.

The system consists of three fluid models and one floating-bodies model. To calculate the

tsunami propagating in the open ocean, inundating land and interacting with structures, two
non-hydrostatic and three-dimensional models are included into the system. The non-hydrostatic model
named STOC-IC to be applied is especially validated through comparison with experimental results and
then applied to an actual port area to estimate damage to the area. Based on these applications it is
confirmed that the non-hydrostatic effects appear in the tsunami overflowing and running up structures
and flowing in the bending part of channel.
The other method is the real-time estimation of coastal tsunami using the inversion method with
offshore observation data of the tsunami. The inversion method is improved by utilizing the location of
the epicenter as a priori information. The results of the numerical experiments confirm that the improved
inversion method can estimate the tsunami source and predict the tsunami profiles along coasts precisely
from the tsunami profiles observed at 5 offshore locations for 15 minutes after the earthquake occurrence.
Especially, the tsunami arrival time and the first tsunami height can be predicted with good accuracy. It is
clarified by the parametric studies how the length of observation time or the number of offshore points
affect the accuracy of prediction.

Acknowledgements
The authors would like to express our gratitude to Geographical Survey Institute Japan for supplying
the airborne laser scanning survey data of the model port. This research is partially supported by the
Ministry of Education, Science, Sports and Culture of Japan, Grant-in-Aid for Scientific Research.

References
Arikawa, T. and T. Yamano (2008): Large-Scale Simulations on Impulsive Wave Pressures by Using
CADMAS-SURF/3D, Annual J. Coastal Eng., JSCE, Vol. 55, pp. 26-30 (in Japanese).
Fujima, K., K. Masamura and C. Goto (2002): Development of the 2d/3d Hybrid Model for Tsunami
Numerical Simulation, Coastal Eng. J., Vol. 44, No. 4, pp. 373-397.
Fujima, H., Y. Shigihara, K. Honda and H. Yanagisawa (2007): Some Detailed Numerical Simulations of
2004 Indian Ocean Tsunami, Proc. 4th Int. Workshop on Coastal Disaster Prevention, pp. 56-64.
Fukuoka, S. and A. Watanabe (1998): Three Dimensional Analysis on Flows in Meandering Compound
Channels, J. JSCE, No. 586/II-42, pp. 39-50 (in Japanese).
Hom-ma, M. (1940): Discharge Coefficient on Overflow Weir (Part-2). JSCE Magazine, Civil
Engineering, Vo. 26, No. 9, pp. 846-862 (in Japanese).
Honda, K. and T. Tomita (2009): Tsunami Inundation Simulation by Three-Dimensional Model. Coastal
Eng. 2008, ASCE, pp. 1433-1445.
Honda, K., T. Tomita, D. Nishimura and A. Sakaguchi (2009): Numerical Modeling of Tsunami Drifted
Bodies, Ann. J. Civil Eng. in the Ocean, JSCE, Vol. 25, pp. 39-44 (in Japanese).
Ikeya, T., R. Asakura, N. Fujii, M. Ohmori, T. Takeda and K. Yanagishima (2005): Experiment on
Tsunami Wave Force Acting on Floating Body and Development of an Evaluation Method, Ann J.
Coastal Eng., JSCE, Vol. 52, pp. 761-765(in Japanese).

- 86 -

Iwase, H., T. Mikami, C. Goto, and K. Fujima (2002): A Comparative Study of Nonlinear Dispersive
Long Wave Equations for Numerical Simulation of Tsunami, Proc JSCE, No. 705, pp. 129-138 (in
Japanese).
Kawata, Y., and N. Koike (1997):. Numerical Tsunami Simulation Based on the Principle of
Superposition, Proc. Coastal Eng., JSCE, Vol. 44, pp.271-275 (in Japanese).
Koike, N (2002): Study on Estimation and Error of Tsunami Initial Displacement of Water Surface Using
Inversion Method with a Priori Information, J. Hydraulic, Coastal, and Environmental Eng., JSCE,
Vol. 712, pp. 117-135 (in Japanese).
Koshimura, S. (2004): Modeling a Tsunami Generated by the Tokaido-Oki Earthquake of Mw 7.3
[04/09/05], http://www.dri.ne.jp/koshimuras/tokaido04/case0/ (online).
Nakatsuji, K., S. Karino, and H. Kurita (1992): Finite Element Analysis of Tidal Flow in the Osaka Bay
with Subgrid Scale Eddy Coefficient, Proc. Hydraulic Eng., JSCE, Vol. 36, pp. 693-696 (in
Japanese).
Nagai, T., T. Kato, N. Moritani, H. Izumi, Y. Terada and M. Mitsui (2006): Offshore Tsunami
Monitoring Network Design Using GPS Buoys and Coastal Onsite Sensors, Proc. 30th Int. Con.
Coastal Eng., ASCE, pp. 1529-1540.
Sakakiyama, T. and R. Kajima (1992): Numerical Simulation of Nonlinear Wave Interacting with
Permeable Breakwaters, Proc. 23rd Int. Conf. Coastal Eng., ASCE, pp. 1517-1530.
Sato, R. (1989): Nihon no Jishin Danso Parameter Handbook, Kajima Institute Publishing Co., Ltd (in
Japanese).
Tanimoto, K., K. Kimura and K. Miyazaki (1988): Study on Stability of Submerged Dike at the Opening
Section of Tsunami Protection Breakwaters, Rep. PHRI, Vol. 27, No. 4, pp. 93-121 (in Japanese).
Tatsumi, D., and T. Tomita (2007): Real-Time Tsunami Prediction Based on Inversion Method Using
Earthquake Information, Proc. 3rd Asian-Pacific Cong. Computational Mech., MS24-2-2.
Tomita, T. and K. Honda (2009): Tsunami Estimation Including Effect of Coastal Structures and
Buildings by 3d Model, Coastal Structures 2007, ASCE, pp. 681-692.
Tomita, T., K. Honda, and T. Kakinuma (2006): Application of Three-Dimensional Tsunami Simulator to
Estimation of Tsunami Behavior around Structures. Coastal Eng. 2006, ASCE, pp. 1677-1688.
Yasuda, T., T. Takayama, K. Kawamura, and H. Mase (2007): Real-Time Tsunami Prediction by Inverse
Analysis of Tsunami Profiles Observed at Offshore Points, Ann. J. Coastal Eng., JSCE, 54, pp.
196-200 (in Japanese).
Yoneyama N., M. Matsuyama and H. Tanaka (2002): Numerical Analysis for Locally High Runup of
1993 Hokkaido Nansei-Oki Seismic Tsunami, J. JSCE, No. 705/II-59, pp. 139-150 (in Japanese).

- 87 -

- 88 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

TSUNAMI EVACUATION SIMULATION

IN AN URBAN AREA OF JAPAN

Kentaro KUMAGAI
National Institute for Land and Infrasracture Management (NILIM), Yokosuka, Japan,
kumagai-k27n@ysk.nilim.go.jp
Abstract
A computer simulation of tsunami evacuation is applied in estimation of the human
suffering due to tsunami, in verification of the effect of the countermeasures assists tsunami
evacuation, and in engaging educational activity for inhabitant. In this paper, major
methods of tsunami evacuation simulation in Japan are classified by three fundamental
viewpoints, such as the estimation method of evacuation start time, modeling method of
evacuee's behavior and evaluation method of the safeness of the evacuee's behavior.
Keywords: Evacuation simulation; evacuation start; evacuee's behavior; safeness
verification

1. Introduction
A computer simulation of tsunami evacuation for residents is applied in estimation of the human
suffering due to tsunami, in verification of the effect of the countermeasures assists tsunami evacuation,
and is utilized for education of inhabitant.
In this paper, major methods of evacuation simulation in Japan are classified by three viewpoints, such
as the estimation method of evacuation start time, modeling method of evacuee's behavior and evaluation
method of the multiple risk of tsunami inundation, destroyed houses and fire. And a case study of tsunami
evacuation simulation was conducted.

2. Tsunami Evacuation Simulation Models

Evacuation start time: To simulate the tsunami evacuation behavior of the evacuee, the first step of the
simulation is to estimate the evacuation start time after the earthquake. The evacuation start time model
can be classified by two characteristic features, modeling method the effect of disaster information and
modeling method of dispersion of the evacuation start time attributed to individual difference.

- 89 -

Table 1 Major models of estimating the tsunami evacuation start time

Fundamental Parameters

Effect of disaster
information

Individual difference

Nagakawa K.
and F. Imamura
(2000)

- Earthquake intensity
- Time of occurrence
- Location of residence
- Tsunami experience

- Communication
channel and impact of
disaster information

No

Hayakawa T.
and F. Imamura
(2002)

- Earthquake intensity
- Time of occurrence
- Location of residence
- Tsunami experience

- Announce time

-Estimated percentage
of evacuee

Modeling method of evacuee's behavior: The second step of the simulation is to model the evacuee's
behavior. Model type can be classified by mainly four characteristics (Research report of Institute for Fire
Safety and Disaster Preparedness, 1983, Imamura, F., T. Suzuki and M.Taniguchi, 2001): Crowd of the
evacuee can be modeled as a fluid body or granular individual agent; Direction of the movement of
evacuee in the street can be restricted only one way flow or allow two way flow; Topography are modeled
into the square mesh, node-link network or without modification; Evacuation routes from the resident to
evacuation site are endogenously-generated in the model or exogenously-generated way.
Table 2 Modeling Method of evacuee's behavior
Crouwd model

Movement

Topography

Evacuation route

Hori M. et al.
(2005)

individual agent

One way

No modification

endogenously

Muti-agent
simulation

Kiyono J. et al.
(1996)

individual agent

One way

No modification

exogenously

DEM

Nishihata T. et
at (2005)

individual agent

One way

Node-Link network

endogenously

Potential Model

Shiraki W. et al
(2006)

individual agent

One way

Node-Link network

endogenously

Cellular
automaton model

Multiple risk: There are risks not only inundation risk but seismic risk such as house wreck or fire
spreading along the tsunami evacuation. Sato, H. et al (2002) includes inundation and house wreck
multiple risks in his simulation. Miyoshi, T. et al (2008) includes fire risks.

3. Case Study of Tsunami Evacuation Simulation

A case study of tsunami evacuation simulation was conducted. It includes house wreck and inundation
multiple risk in the tsunami evacuation simulation.

References
Hayakawa, T. and F. Imamura (2002): Model of the Determination of Tsunami Evacuation and Its
Application, Journal of Japan Society for Natural Disaster Science, Vol. 21-1, pp. 51-66.
Hori, M., Y. Inukai, K. Oguni and T. Ichimura (2005): Study on Developing Simulation Method for
- 90 -

Prediction of Evacuation Processes after Earthquake, Sociotechnica, Vol.3, pp. 138-145.

Imamura, F. and T. Suzuki and M. Taniguchi (2001): Development of a Simulation Method for the
Evacuation from the Tsunami and Its Application to Aonae, Okushiri Is., Hokkaido, Journal of Japan
Society for Natural Disaster Science, Vol. 20-2, pp. 183-195.
Institute for Fire Safety and Disaster Preparedness (1983): Report of Evacuation Simulation Research, pp.
3-4.
Kiyono, J., F. Miura and K. Takimoto (1996): Simulation of Emergency Evacuation Behavior during
Disaster by Using Distinct Element Method, Journal of Japan Society of Civil Engineers, No.537/
I-35,pp. 233-244.
Miyoshi T., Y. Matsushita, W. Hashimoto, E. Itoigawa and O. Murao (2008): Wide Area Evacuation Plan
based on Fire Incidents and Fire Spreading Risk, Journal of Social Safety Science, No.10, pp.
105-110.
Nishihata, T., Y. Moriya, T. Tamura, K. Takimoto and F. Miura (2005): Simulation of Tsunami Evacuation
at the Bay Kesennuma, Annual Journal of Civil Engineering in the Ocean, JSCE, Vol. 21, pp.
163-168.
Shiraki W., H. Itsura and H. Aritomo (2006): Developing the Participatory Type Evacuation Simulation
System for Live Design, Proceedings of Annual Meetings of JSCE, Vol.61, pp. 379-380.

- 91 -

- 92 -

Ports and Harbors Bureau, MLIT, Japan

Tsunami disaster reduction in

Japanese ports and harbors

2nd December, 2009

Naokazu Takata
Ports and Harbors Bureau,
Ministry of Land, Infrastructure, Transport and Tourism, Japan

Todays Topics
Part 1: Coast Administration in Japan
Part 2: Tsunami disaster reduction in
Japanese ports and harbors

- 93 -

Part 1: Coast Administration

in Japan
1. Coastal Disasters in Japan
2. Coast Administration in Japan

Coastal Disasters in Japan

Area that can be easily damaged by Storm Surges
Area that can be highly damaged by Coastal Erosion
Path of typhoon
Tsunami
Winds and waves in winter season

Hokkaido Southwest
Offshore Earthquake (1993)
M = 7.8

Area where ground can subside significantly

Epicenter
Hyogo Prefecture Southern
Part Earthquake (1995) M = 7.2

Sea of Japan Central

District Earthquake
(1983) M = 7.7

Kushiro Offshore
Earthquake (1993) M = 7.8

Tokachi Offshore
Earthquake (1968) M = 7.9

Sanriku Offshore
Earthquake (1933) M = 8.1
Sanriku Offshore
Earthquake (1896) M = 8.5

Ansei Tokai Offshore Earthquake (1854)

Damage caused by
Typhoon 9 (1958)
Damage caused by
Typhoon 19 (1991)
Damage caused
by Typhoon Ruth
Damage caused by
(1951)
Typhoon 18 (1999)

Damage caused by Typhoon Kitty (1949)

Damage caused
Damage Damage caused by Typhoon Kanogawa (1958)
by Typhoon
caused by
Isewan (1959)
Typhoon 26 (1966)
Tonankai Earthquake
Damage caused Damage
(1944) M = 7.9
by Typhoon caused by
Muroto (1934) Typhoon
Nankaido Earthquake
13 (1953) (1946) M = 8.0

- 94 -

Storm Surges and Coastal Erosion

Major Storm Surges in 2004

Typhoon No.16 in Aug.

Typhoon No.18 in Sep. Typhoon No.23 in Oct.

Tide Level, 4.36

Takamatsu Record

Wind Velocity, 60.2m/s

Hiroshima Record

Wave Height at Muroto, 13.5m

Domestic Record

46,625 houses were flooded in

Japan

21,000 houses were flooded

in Japan

13 houses collapsed and 3

people were killed.

Kobe City

Takamatsu City

Muroto City

Severe Winds and Waves in Winter Season Causing Coastal Erosion

History of Earthquakes & Tsunamis in Japan

Major Earthquakes & Tsunamis Since 1896
Hokkaido Nanseioki M

Meiji Sanriku M

July 12,

Dead or Missing:
Max Tsunami Height:

June 15,
Dead:
Max Tsunami Height:

Showa Sanriku M

Nihonkai Chubu M
May 26,

Dead:
Max Tsunami Height:

January 17,

Dead or Missing:

Dead or Missing:

Mikawa M
January 13,

Dead:

Tonankai M
Showa Nankai M

Dead or Missing:
Max Tsunami Height:

September 1,

March 7,

Hyogoken Nanbu
M

Kanto M

Kita Tango M
Dead:

March 3,

December 7,

Dead:

December 21,

Dead or Missing:
Max Tsunami Height: m

Chile M
May 23,

Dead or Missing:
Max Tsunami Height:

- 95 -

Classification of Japanese Coast

Length of Coast
35,000km
MAFF1,700km
FA3,400km

Protection
Area

RB5,500km

15,200km
PHB4,400km

Constructing
Coastal
Protection
Facilities
Requiring
Permission for
Exclusive Use
Restricting
Activities

Area Regulated
by Coast Law

MAFF&RB200km

Other Public
Coast

Requiring
Permission for
Exclusive Use
Restricting
Activities

Public Land
8,500km

8,500km
Road, Railway,
Private Land, etc

MAFF
FA
RB
PHB

12,500km

Ministry of Agriculture, Forestry, and Fisheries

Fisheries Agency
River Bureau, MILT
Ports & Harbors Bureau, MILT
6

Characteristics of Coast
in Ports & Harbors Bureau Jurisdiction
Density of Population & Assets is high in the hinterlands of coasts in Ports
& Harbors Bureau Jurisdiction.

Tokyo

Kure

Osaka

Size of Protection Area

Length
PHB Jurisdiction

Other Jurisdiction

Port Located
Municipalities

Population
PHB Jurisdiction

Chiba

Other Municipalities

Industrial Sales
Port Located
Municipalities

Other Jurisdiction

Other Municipalities

- 96 -

Structure of Coastal Administration

National Government
Ports and Harbors Bureau,
Ministry of Land, Infrastructure
and Transport

Port and Airport

Research Institute

Setting a basic policy

National Government Works Project
Construction, Major Improvement etc

Local Government

Research and development

Maintenance
Subsidized Project
Construction, Major Improvement,
Amenity Oriented etc

System of Coastal Works

Share of cost of coastal works between national government and local government
National Government Local government

Construction,
Major Improvement, etc

Amenity Oriented
(Beach, Green Belt, etc)

National Government Works Project

Subsidized
Project

Maintenance
(Seawall)

(Floodgate)

Hiranai

(Beach, Green Belt)

Tokyo Port

Hakata Port
9

- 97 -

Part 2: Tsunami disaster reduction

in Japanese ports and
harbors
1. Risks of tsunami disaster in Japan
2. Tsunami countermeasures by
Ports & Harbors Bureau

10

Tsunami Protection Committee

Circumstances
December, 2004
Indian ocean earthquake and
tsunami

February, 2005
Establishment of Tsunami protection
committee

March, 2005
Recommendations of the tsunami
protection committee

Summary of the recommendation

The first recommendation of comprehensive tsunami
protection in Japan
1Basic policy for tsunami protection measures in the coming
years
zStrategically and strongly implement an integrated combination of
structural and nonstructural measures as comprehensive disaster
mitigation measures.
zEstablish comprehensive pre-/post- tsunami measures through the
allocation of the roles of "self-help," "mutual assistance" and "public
assistance.
zIncrease public awareness of the importance of tsunami preparedness.

2Specific urgent goals and damage mitigation measures

zWarning and information provision
zPreventive measures(evacuation, tsunami protection facilities, better
land use)
zPost-tsunami measures(damage information, transportation networks,
restoration/rehabilitation)
zAccumulation and dissemination of technology and knowledge for
tsunami disasterprevention

3Medium to long range goals and tsunami protection

measures

Indian ocean tsunami

zMedium to long range tsunami protection measures including

emergency measures .
zMeasures that take demographics into consideration.
zMeasures against sea level rise due to global warming.

11

- 98 -

Four tectonic plates around Japan

North American Plate
Eurasian Plate

Pacific Plate

Philippine Sea Plate

12

Imminence of Big Earthquakes in Japan

NEMURO

Probability of a big earthquake

which can occur in 30 years

SANRIKU

MIYAGI

HYUGA

Southern KANTO

TOKAI
(Reference)
TONANKAI

NANKAI

13

- 99 -

Estimated height of tsunamim

Estimated Height of Tsunami

25
20
15
10
5
0

Tonankai & Nankai

Earthquake

Tokai
Earthquake

Chishimakaikou

&
Nihonkaikou

Earthquake

14

Systems of Tsunami Countermeaures

in Coast Administration
Urgent

Steady

Promotion of
Tsunami Hazard Map

Repair of
Coastal
Protection Facilities

Facilitation of
Evacuation

Automation & Remote

Control of Gates

For Lives

Construction of
Coastal
Protection Facilities

For Assets

15

- 100 -

Coastal Protection Facilities

Seawall

Miyako (Iwate Pref.)

Tsunami Breakwater
Constructed
Under Construction

Tsunami
Breakwater

Suzaki Port (Kochi Pref.)

16

Seawalls with sufficient crest height

Some seawalls do not
have sufficient crest height
to
prevent
tsunami
disaster.
Insufficient
16.7

Not investigated
23.9

Tsunami overtopping seawall in Iwate

(Chile Earthquake in 1960)

Sufficient
59.5

Fig. Crest height status shown by the ratio of

coastal protection facility length (total=14,684 km)

- 101 -

17

Earthquake-proofing in Construction
Some
seawalls
could
be
damaged by a large earthquake.
Not earthquake-proofed
11.4

Earthquake-caused structural damage

Earthquake-proofed
34.3

Not investigated
54.3

Fig. Status of earthquake-proofing shown by the ratio

of coastal protection facility length (total=9,437 km)

Anti-liquefaction work

18

Automation and Remote Control of Gates

Some gates might not be
closed before a tsunami
strikes.
Not closed in time
21.6

Closed in time
54.5

Not investigated
23.9

Fig. Ability of gate to close in time shown by

the ratio of gate numbers (total=7,685 gates)

- 102 -

19

Tsunami Prevention Station

Meteorological
Agency

Remote
Gate Control

Gates

Meteorological
Information

Tsunami Prevention
Station
Information

Wave/Sea Level
Observation

Evacuation

20

Disseminating Hazard Maps for Tsunami

Many local governments
have not yet made a
hazard map.

Already made
319 governments
Not yet

Suzaki City (Kochi Pref.)

49

330 governments
51

Manual For Hazard Maps

Promotion of Mapping Support

Fig. Status of hazard map preparation by

local governments (total=649 governments)

Disseminating Hazard Maps

21

- 103 -

Comprehensive tsunami response drills

Ministry of Land, Infrastructure and Transport conducted

comprehensive tsunami response drills2009.7.4
The number of participating institution : 51(The Japan
Coast Guard, The Self-Defense Forces, Chubu Regional
Development Bureau, Shizuoka Pref., Shizuoka City etc)
The number of participant : about 10,000

Drill on land
Drill on the sea

Main site (Shimizu port, Shizuoka Pref.)

Collecting information drill

Rescue work drill

Clearance of obstacles

Investigation of facilities

Withdrawal of oil

Transportation of
support goods

Patrol boat

Jet

Transportation of victims

22

Offshore Observation of Tsunami

Offshore observation of tsunami
is necessary so that warnings
may be issued appropriately.
GPS Tsunami Observation System

Pic. GPS Buoy Tsunami Meter

installed 13km off the Muroto-Misaki
by PARI, Tokyo Univ. and Hitachi
Zosen Corp.

23

- 104 -

Observation by Tsunami Buoy

Tokaido-off-earthquake Tsunami, Sep.5th 2004
GPS Buoy
0:31 10cm

13km from Tide

Station

Tide station

0:40 32cm

24

Comprehensive Tsunami Countermeasure

Meteorological Agency

Information

Information

Base Station

Coast
Guard

Evacuation
Facilities,

Fence etc.

Hazard
Map

Automatic
Warning
1.5

Breakwater

Improvement

Elimination of

un-managed
boats

Warning

Ports and
Harbors Bureau

TSUNAMI

Automatic

Flood Gate

GPS
GPS Tsunami Buoy

Check,
Study

for Facilities

Coastal Protection

Facilities

Hard
Approach

Soft
Approach

- 105 -

25

Thank You Very Much

- 106 -

Proceedings of the 6th International Workshop on

Coastal Disaster Prevention, Bangkok, Thailand,
December 1-2, 2009

Recent Progress on Tsunami Disaster Mitigation in Indonesia

Subandono DIPOSAPTONO
Director for Coastal and Ocean Affairs, Ministry of Marine Affairs and Fisheries, Jakarta Pusat, Indonesia,
subandono.diposaptono@yahoo.com

Enggar SADTOPO
Head of Coastal Disaster Mitigation Division, Ministry of Marine Affairs and Fisheries, Jakarta Pusat,
Indonesia

Abstract
Coastal areas in Indonesia are highly vulnerable to tsunami hazards that have considerable
impacts to human life and the environment. In this regards, the need for coastal disaster
reduction in Indonesia is growing along with the growing development and increasing
population in these areas. Many activities have been done to assess tsunami potentials,
human and natural vulnerabilities, and mitigation effort to reduce its impacts. The disaster
mitigation in this case should be done in all levels and by all related stakeholders.
Governance, policies and regulation are made to give directions and provide legal assurance.
Structural and non-structural countermeasures are done to strengthen the society social,
economical and environmental conditions.

1. Background
Indonesian coastal areas has developed along with the increased of population and cities since long
time ago. Coastal area provides space, ways for transportation and access for distribution of people and
goods that eventually accelerate the development of trading, settlement and city. Coastal area with white
sandy beaches becomes a beautiful place for recreation and tourism. Almost in every Indonesian main
island, we can find many big cities growing in coastal area that is now very developed and populated.
The problems arise as these coastal areas are also very prone to natural hazards such as erosion, floods,
cyclone, earthquake, and tsunami. Whether we are aware or not, we put ourselves in more risks and
danger by developing more infrastructures, economy landmarks, and settlements in coastal areas.
Tsunami is one of the most catastrophic natural hazards in the Indonesian coastal areas. Usually, these
tsunamis are very devastating and have considerable impacts, the highest ever impact was the Indian
Ocean Tsunami in 2004 and the South Java Tsunami in 2006.
This paper elaborate recent progress on tsunami disaster mitigation efforts done by stakeholder in
Indonesia both for the structural and non-structural countermeasures

- 107 -

2. The Nature of Tsunami Disasters in Indonesia

Indonesia is located at the convergence of three main tectonic plates: the Eurasian, Indo-Australian and
Pacific Plates. The Indo-Australian plate sub-ducts beneath the Eurasian plate along the Sunda Arch as
part of a greater arc syste Subandono Diposaptono m in the Indian Ocean. Beyond the Sunda Arch, the
Banda sector continues eastward, where the oceanic arc collides with the Indo-Australian plate.
All around Indonesia are sub-duction zones where the large scale, slow cycle of land is driven by the
rivers flowing down mountains coupled with the deep Earth heat driving the plate movements from below.
On the perimeter, converging plates slowly store energy resulting it many small, shallow seismic events
and the occasional deep earth, high energy event. The concentrated Earth energy flows (sun, tide and
deep earth heat) converge on the Islands of Indonesia resulting in much active land building especially by
volcanic activity.
The sudden release of large amounts of stored energy (as friction) where two plates shift across a broad
area that has caused the catastrophic Aceh Tsunami on 26 December 2004. This giant tsunami was the
result of the high intensity, sub-sea Sumatra Earthquake that impacted coastal areas not only in NAD
(Nangroe Aceh Darussalam) and North Sumatra provinces of Indonesia but also as far a field as Malaysia,
Thailand, Sri Lanka, India, Maldives, and Africa.
Other examples of tsunami disasters in the last two decades in Indonesia include Flores (1992) with
more than 1950 dead, East Java (1994) with 240 reported deaths, Palu (1996) with 3 lives lost, Biak
(1996) with 107 reported deaths, Banggai (2000) with 4 deaths, West Java (2006) with 668 deaths, and
Bengkulu (2007) (no data on losses)1.
Some of the Indonesian coastal areas have a high potential risk to tsunami i.e. the West coast of
Sumatra, South coast of Java, South coast of Bali, North and South coast of Nusa Tenggara, islands of
Maluku, North coast of Papua, and most of Sulawesi (Celebes) coast (Fig. 1).
Indonesia has been affected by Tsunami since a long time is history. There are records of more than
100 such events over the last 400 years. These records indicate that between 1600 and 2007 there have
been 109 tsunamis. This suggests that the energy of earthquakes and subsequent tsunamis increases with
time interval between earthquakes and recorded tsunami of approximately once every four years. The
data suggests that frequency has increased in the last half century although it is not certain if this is
because of better records and monitoring or reflects greater seismic activity. Certainly the potential for
impact of coastal communities has increased greatly in that time. From 1960 - 2007 there have been 22
significant tsunamis. This indicates that the frequency of tsunamis is around one in every two years2.
Table 1. shows tsunami cases in Indonesia during 1960-2007.
While the speed and scale of the 2004 tsunami was one main reason for the loss of life and property;
non-disaster focused land use planning, lack of awareness, lack of infrastructure for tsunami prevention,
no warning system, and the deterioration of the coastal environment also played their part. In the worst
hit Aceh region, many buildings were found to have been constructed without compliance to the relevant
sections of the building code for earthquakes and tsunami3. Other implicating factors included: houses

ITDB, 2004 and local reports.

Number of earthquake events may decline logarithmically with increased intensity of the event. Time
interval increases correspondingly. Tsunamis probably also follow this pattern.
3
Unfortunately this may be representative of Indonesian coastal areas in general.
2

- 108 -

built very close to the sea; no greenbelts, and only remnants of the original mangroves and coastal forests
remaining.

Eurasian Plate

Pacific Plate

BANDA ARCH

SUNDA ARCH
--- ---

Potential Tsunami
Indo-Australian Plate
Figure 1. Potential Tsunami in Indonesia

Tsunami disasters in Indonesia caused widespread damage and suffering as the ensuing Tsunami wave
impacted on the coastal zone. An estimated 300,000 lives were lost. The tsunami run-ups were reported
higher than 30 m. As Aceh is relatively close to the Earthquake epi-centre off the coast of Sumatra and
the velocity of the Tsunami wave was very fast, there was little time for alert and evacuation. In the wake
of the soul searching in the months after the catastrophe it was concluded that although Indonesians are
now more aware of the danger posed to population centers located in coastal areas more needs to be done
to pass on information about Tsunamis and how one may more away from the main danger area.

3. Tsunami Disaster Mitigation

The 2004 Indian Ocean Tsunami has pushed tsunami disaster mitigation action progress by different
institution significantly forward. Moreover, institution has realized the importance of coordination
between these institutions to have more leverage in the progress output.
The priority issues for tsunami disaster mitigation include nine intervention areas: i) legislation and
regulation, ii) integrated coastal planning and management, iii) coastal disaster mitigation plan, iv) coastal
community empowerment, v) coastal habitat rehabilitation, vi) retrofitting, vii) risk assessment, viii)
development of evacuation route, and ix) tsunami early warning system.
3.1 Legislation and regulation
The recent developments in Indonesia have been forcing the new awareness and direction in disaster
reduction initiatives. The enactment of Disaster Management (Law No. 24/2007) and Coastal and Small

- 109 -

Island Management (Law No. 27/2007) provide Indonesia a strong legal basis for better disaster reduction
program.
The National Disaster Management Agency was formalized through Law Nr. 24/2007 and has a
coordinating role. According to this law, disaster reduction should be institutionalized through national
and local development plan. This perspective will ensure the sustainability of the program in term of
planning and budgeting. This also will reduce the inconsistency of the policy due to the change of
national and local leader.
The Law No. 27/2007 gives clear mandate for all stakeholders in reducing the negative impacts of
coastal disaster and increasing community participation and environmental conditions. The law 27/2007
absolutely gives a legal instrument to mitigate coastal hazards. The law 24/2007 also provides a direction
for better disaster governance. Law No. 27/2007 law has been translated in a more operational regulation
i.e. Government Regulation on Coastal Zone and Small Island Disaster Mitigation.
3.2 Integrated coastal planning and management
To reduce the impact of coastal disasters in Indonesia, nationally we always improve our capabilities to
mitigate these events. The Ministry of Marine Affairs and Fisheries (MoMAF), Republic of Indonesia is
also pro active in minimizing the impact of coastal disaster on coastal communities and on aquaculture
activities. The program emphasized the implementation of Integrated Coastal Zone Management (ICM).
The ICM will try to balances the utilization between natural resources, human needs, and disaster
mitigation aspects. The program have been implemented in 15 provinces in Indonesia and now started to
be replicated in other provinces.
Those three aspects are important to be noted because they influence and are related to each other that
eventually produce the coastal socio-ecology resilience. A healthy coastal ecosystem condition guarantees
that sustainability of exploitation activities by humans. For example, the health of mangroves and corals
affects the productivity of waters and fisheries, also effects under water tourism and building in the
coastal area from the wave attacks. On the contrary, exploitation of area and resources in the coastal area
will affect the balance and health of the environment due to waste, physical construction, or the profile
changes of the coast. And lastly, ecological condition, due to the exploitation by humans will be lost or
deter if there are no disaster mitigation concept in the coastal area.
Within this approach the hazard respond has changed from single agency to partnership and proactive
measure by assessing vulnerability to improve the risk management that developed with communities.
Accordingly, human activities in coastal area should not degrade and reduce the environment capacity to
absorb the hazard, increase human vulnerability by developing building and infrastructure that cannot
defend from hazard, and reduce their capacity to cope with and recover from hazard.
3.3 Disaster Mitigation Plan
According to ICM planning, the disaster mitigation plan is develop to give more emphasis on the
disaster reduction program. This mitigation plan reflects the change in disaster management paradigm in
Indonesia. It is reflected in more community planning such as land use planning, inundation management
plan, reduce vulnerability, and increase coastal community resilient. An integrated hazard mitigation plan
is important especially in very developed coastal area. This plan provides guidance for stakeholders in
reducing their vulnerability to coastal hazard for long time. Hazard mitigation or management plan is

- 110 -

directed to reduce the human vulnerability in all possible aspects. It covers activities to reduce the
weakness aspects and improve access to resource, social integration, institutional coordination, public
awareness, and building safety. The plan must be discussed and communicated with another area because
the hazard is not limited by administrative boundary.
MoMaF have been facilitating 4 local governments i.e. Padang, West Sumatra, Serang, West Java,
Denpasar Bali, and Lombok Tengah, West Nusa Tenggara in developing participatory local strategic plan
for coastal hazard mitigation. By doing this, it does not mean that the impact of coastal disaster can be
avoided, however, if planned accordingly, with good education and campaign in the community,
mitigation will eventually reduce the loss of lives, social, and economic and also improve the level of
resilience for future event.
As a final remark, it should be given a word about how to come up with holistic mitigation plan. The
key work is participatory strategic planning. Strategic planning is a planning method used by businesses,
governments, and NGOs to reach goals and get better results. A strategic plan consists of the following
element:
1) Identifies needs,
2) Clarifies goals,
3) Sets detailed objectives,
4) Establishes an action plan & timelines,
5) Provides for periodic progress review
By adopting strategic planning procedure for disaster strategic mitigation planning, it becomes easier to
incorporate holistic mitigation framework by setting priorities among economic, physical, and life.
In order to increase the compliance with the resulted strategic plan, many theories and practice
recommend soliciting participation of major stakeholders in strategic planning process. Workshops can be
a very effective tool for participation. With proper introduction of information and time pressure, group
of people participated in workshop can reach logical and reasonable conclusions, and they would show a
high commitment for their conclusions.
Participatory strategic planning process is really a capacity building process among stakeholders. For
any disaster mitigation, collaboration and cooperation among stakeholders are the indispensable element
for success. Participatory strategic planning process allows all stakeholders to be involved from the
beginning and avoid arbitrary imposition of top-down solutions. Along the planning process, proper
information of all aspects of disaster management would be provided so that the participants could enrich
their knowledge and deepen their understanding.
3.4 Coastal community empowerment
The empowerment in this context includes the improvement of economic livelihood, social
cohesiveness, community awareness, and access to capital and market. Those things area important in
enabling coastal community to prepare for disaster, recovery from event, and adapt to change that make
them more resilience in the future disaster.
In this regard, public education must be conducted regularly. Issues such as nature of hazard,
probability and magnitude, area likely to be flooded, proper responds and community preparation are
necessary to know. Picture, map, questioner, event scenario are useful in awareness and education

- 111 -

program.
The Ministry of Marine and Fisheries also use traditional and cultural events in the awareness
campaign. This method is intended to increase community resilience that should understand the nature of
hazard, able to mitigate the impact, disseminate and change information and have a disaster mitigation
planning.
In collaboration with Sea Grant Program USAID under Indian Ocean Tsunami Warning System
(IOTWS) we conduct training in coastal community resilience. This training intends to increase the
community understanding on their risk to coastal disaster and how they could increase their resilience to
cope with that event. The program covers activities such as: i) identification of issue, ii) conduct an
assessment on coastal community resilience for small community, iii) develop action plan to increase
community resilience.
The Indonesia Research Agency/community preparedness program provides outreach, education and
campaign material and implements community awareness programs to schools and community all over
Indonesia.
Local and Internationa NGOs has contributed much especially on community education and awareness.
Kogami, a local NGO in Padang provide education to the community on earthquake and tsunami, train
community on tsunami evacuation, and policy advocacy for disaster management in Padang and other
area. Here Surfaid provides community awareness and education programs and funds for tsunami
community preparedness for earthquake and tsunami for Padang and Mentawai and other areas in West
Sumatera. Yayasan IDEP, a local NGO based in Bali, provide community preparedness toolkit for
awareness raising for various hazards including tsunamis. MPBI, has advocated the passing of Law
Nr.24/2007 on Disaster Management and provide community preparedness books, brochure, posters, etc
for various hazards including tsunami
3.5 Coastal habitat rehabilitation
The objective of the habitat rehabilitation is to increase the coastal environment capacity to provide its
services for livelihood and protection from coastal hazards. It has been implementing in the form of
coastal forest and mangrove planting, coral reef transplantation, stock enhancement, and conservation.
Although Indonesia has suffered from a large number of tsunami events since ancient times, until now
there were few effective countermeasures against them. Tsunami countermeasures using hard structures
(seawalls, breakwaters, etc), require on-going maintenance and replacement perhaps every 50 to 100
years and present major infrastructure investment challenges for Indonesia. In general, these initiatives
may not be a good match in Indonesia for the following reasons:

these structures may be unpopular for people living on the coast as they may restrict movement
and social relationships,

tsunami countermeasure using hard structures raise aesthetical and environmental problems and

they can interfere with terrestrial-marine ecosystem function.

It may be that better information, restoring key areas and adapting to the coastal zone is more realistic.
Coastal areas and their economies are often highly productive as the activities are supported by the
renewable inputs from the sun, rivers and its organic matter, sediments and chemical potential, tides,

- 112 -

winds and especially waves. But often economic developments in the fossil fuel era diverted the original
values of the necessary support systems such as clearing much coastal forests because their contribution
was misunderstood. This was poor public policy because in the longer term, fisheries and aquaculture
have declined, wastes have accumulated, tourism has moved elsewhere, disasters impact more lives and
general well being is lower.
Learning from the above institutional weaknesses, and using principles of energy and self organisation
of eco-systems, naturally stabilizing structures such as coastal forests (greenbelt) may be a better
approach. Coastal forests are considered important in reducing the impacts of the Indian Ocean tsunami
based on preliminary environmental assessment, anecdotal evidence, and satellite photography before and
after the tsunami event. For example, in the Aceh coastal area, houses protected by coastal forests
suffered little damage compared to those bearing the full impact of the tsunami waves.
In addition to this potential for tsunami impact mitigation, coastal forests also have environmental
values such as storing Carbon and for wildlife habitat. There are also economic incentives: they are
largely self-organizing after being established (require little maintenance and last a long time) they reduce
nutrient enrichment of the coastal waters and improve the image of the area (important for tourism and
locals alike).
Many coastal areas in Indonesia prone to tsunamis are naturally high energy areas (see above).
Because wind and wave energy inputs are usually highest, they are often dominated by beaches and sand
and are generally not suitable as mangrove habitats4. The alternative for coasts may be to re-establish
them with coastal forests (local names are waru, ketapang, cemara).
A major purpose of the coastal forest is to reduce the tsunami force and its collateral effects landward.
As was seen after Aceh, the seaward coastal forested area, although impacted by tsunami waves or by
other hard debris borne by the waves, served to disperse the wave energy perhaps exponentially with
distance inland and so generally buffered the forested areas behind. In this way, tsunami waves and their
impacts may be reduced very significantly in a short distance and before reaching urban areas. Coastal
forest ecosystems in general also stabilize dunes in the beach zone. More sand retention on beaches also
serves to protect coastal areas from impacts from Tsunamis.
3.5.1. Coastal Forest Effects on Reducing Tsunami impact
Structural counter measures using natural protection methods (soft structure) such as coastal forests
against tsunamis are currently being developed by MoMAF. Activities include planting of Casuarinas,
Hibiscus tiliaceus, and Terminalia cattapa. The sites are located in tsunami prone coastal areas. The
overall goal of this effort is to establish enough coastal forest to significantly reduce the scale of tsunami
impacts. The degree or success of coastal forest in fulfilling this role is not, however, straightforward and
depends on a number of factors, e.g. the distance from the tsunami, the spatial extent and structure of
coastal forest which in turn are dependant of soil conditions, environmental and economic factors such as
available land and conflicting uses.
Although Indonesia has extensive and diverse coastal vegetation, technical research into the
engineering aspects and function of coastal vegetation in protecting coastal areas is still in its early stages.
Most of the research has been on mangrove effectiveness in reducing tsunami impacts in physical models
4

Locations suitable for mangroves are generally limited to areas where rivers flows and mixing, chemical
potential and tides are high such as river mouths, deltas and estuaries.

- 113 -

see (Latief H., 2000; BPDP-BPPT, 2003, and Widjo K, 2004).

Unfortunately, as noted above, many coastal areas in Indonesia prone to tsunamis are not suitable for
mangroves. Furthermore, research on coastal vegetation function other than mangroves is much less.5 In
addition, little hard data or guidance for engineering standards on use of coastal forest for dissipating
tsunami energy are available. Therefore, the following quantitative evaluations coastal forest
effectiveness on tsunami impacts are based on the limited experimental and numerical simulations by
Harada and Imamura (2003).
Harada and Imamura conducted research on a coastal forests effectiveness in reducing tsunami forces
in 20026. Table.1 shows the reduction rate of maximum run up distance, inundation depth, current and
hydraulic force just behind the forest. To simulate mature forest conditions, forest density and diameter of
trunks were selected using 30 trees/100m2 and 0.15m.
The results of the Matrix show the run up distance and the inundation depth decline with increase of
forest width and tsunami height. The current velocity and the hydraulic force also decrease with increased
forest width and tsunami height. Moreover, the rate of reduction of the current velocity and hydraulic
force are greater than for run up distance and inundation depth. The reduction effect on hydraulic force is
greatest overall. This indicates that forests dissipate tsunami kinetic energy more effectively than for
other impacts. As this is probably a function of their structure, these results are not unanticipated.
3.5.2. Application in the Field
The tsunami height for the design of control forest is based on historic maximum heights recorded or
reported in the past and obtained from the historical documents and data base (1992 Flores Tsunami, 1994
East Java Tsunami, 2006 West Java Tsunami, and 2007 Bengkulu Tsunami). For locations where
historical data are not available, the tsunami height is estimated by quantitative simulations and by
correlating bathymetric data from the ocean and coast with past earthquakes. The seismic gap is used to
predict future potential impact zones. The numerical code of TIME (Tsunami Inundation Modeling
Exchange) was applied in this study. TIME is an internationally accepted methodology to simulate
tsunami inundation and its arrival time developed by DCRC (Disaster control Research Center) of
Tohoku University under the support from Japan and United Nations Intergovernmental Oceanographic
Commission (IOC).
Based on tsunami height for the design of the control forest (tsunami height was selected to be 3 m),
research data on effectiveness of control forests and the Law 27/2007 on coastal setback distance, the
coastal forest dimension. The width of coastal forest is 100 m with the density 30 trees/100m2 and using
Casuarina spp (Coastal Casuarina), Terminalia catappa, and Hibiscus tiliaceus. The use of these species is
based on local availability and suitability for beach substrates that are predominately sand.
3.6 Retrofitting
Indonesia has a rich traditional architecture adapted to the climate, social organization and extreme
natural events that are part of the geo-physical reality of Indonesia. One of the basic principles of
5

see for example Harada and Imamura (2003). The quantitative and concrete functions of non
mangrove (coastal forest) to reduce a tsunami in Indonesia are not yet established and
formulated.

from Tohoku University, Japan.

- 114 -

architectural design is that it has to be responsive to the condition of the nature include infrequent natural
disasters. Traditional house is usually constructed in form of stilt house.
In the longer time frame of generations, Indonesias architectural design was adapted to the conditions
that are a feature of the region. In more recent times, however, much of this tradition has been ignored
with the result that when an earthquake and/or tsunami strikes, many buildings are destroyed and many
lives unnecessarily lost.
Based on the past experience, MoMAF has retrofitting houses in several coastal areas in Indonesia.
Some of the basic concepts are as follows: The first floor is two meters high from the ground so when the
sea water level rises, the water would only pass but not enter into the house. To make it stronger, the
footplate concrete foundation and the reinforced concrete column uses a reinforced concrete column with
diameter of 30 cm. This main post or column functions to deflect the impact of debris that is carried by
the tsunami. This debris can be quite large, for example, boats and tree trunks. The column is cylindrical
and of smaller diameter to deflect the impact caused by debris. The foundation is set to a depth of 1.5 m
(or to the level of the bedrock) to avoid the scouring by tsunami.
Even though the typology of the buildings used is the same, the form and construction differs from
each area. The houses and construction is adapted to the structure and characteristics of the soil.
Besides proper construction techniques, the use of building materials is being modified. In the planning
of the house, light weight materials are being used to reduce the weight of the building. Light bricks are
used for the wall or wooden paneling/plywood and zinc sheeting is used for the ceiling rather than heavy
clay roof tiles. The floor uses wooden boards. Only the floor in the kitchen uses reinforced concrete to
avoid fires. The MoMAF has also re-formulated the building code for earthquake and tsunami to include
the existing traditional design with some modern building techniques learnt from experience in
earthquake and tsunami friendly house construction.
The recently earthquake in Padang, Pariaman in West Sumatera (7,6 richter scale) on September 30,
2009 and in Jambi (7,0 richter scale) the next day has created economical losses of more than 2 billion
Rupiahs. The highest affect areas are Padang, Pariaman, Agam, Pasaman, and West Pasaman. As
observed, many houses were devastated in the event. This is due to those houses were not engineered
built. They did not have practical columns, ring balk and sloofs. The quality of the materials was low and
other imperfection.
Until 2008, MoMAF has built 497 houses in 18 Provinces. In Pariaman and Agam, these houses were
proven to be earthquake resistance. In other southern coastal areas, these engineered houses could stand
the Bengkulu earthquake of 7,4 RS and a ship of 10 GT pushes by a 1-2 meter tsunami.
In 2009, 2.236 disaster friendly fishermen houses were built in 55 cities. There are two types of houses
being built, stilt and non-stilt houses. The Stilt houses were built to let tsunami pass under. In normal
situation, the lower ground can be used for leisure location, parking place, and other purposes. The upper
level can be an evacuation place when a tsunami strikes.
3.7 Tsunami Early warning System
Early warning saves lives. Thats a very obvious lesson from the tsunami event. Early warning
systems are considered the foundation of disaster mitigation.

With the advances in science and

technology, accurate forecasting of the occurrence of a natural hazard has saved thousands of lives and
protected properties. It is very unfortunate indeed that the Indian Ocean lacks a tsunami warning system

- 115 -

like the one installed in the Pacific.

Indonesia is prepared and committed to develop and manage a National Tsunami Early Warning
System (TEWS) as part of the Regional Indian Ocean TEWS. Indonesia has designed framework that will
be used for establishing an effective and durable Indonesian TEWS on national scale which could provide
a tsunami warning in less than 5 minutes wherever earthquakes occur. The numbers of TEWS are 22
buoys which consist of 2 buoys from USA, 9 buoys from Germany, and 11 buoys from Indonesia. This is
coordinated under the Ministry of Research and Technology and activities are done together with other
related stakeholder for other government agencies, international agencies and local NGOs. GermanIndonesia TEWS and USAID under the Indonesian Ocean Tsunami Warning Systems has provided
capacity building in this regards.
The high technology early warning system proposed7 is, however, expensive, is subject to tampering
and/or theft and requires on-going maintenance and possibly has safety issues8. Although it can be a
useful tool to warn of tsunamis, this is only one aspect of the issue. In addition to warning, people in
harms way need also to:

understand what is happening,

how to respond and

what to expect in the aftermath.

Another challenge to overcome is the time imperative for issuing tsunami warning. Even in the event
that an early warning system is deployed, Indonesia is a vast Archipelago with over 3,000 inhabited
islands with all the concomitant communication challenges that this implies.

Even with better

communication it may not be possible to adequately warn all people in all potential impact areas in a
timely fashion. In conclusion, this system has a role to play in Indonesia but to be more effective,
additional information transmission methods about Tsunami need to be developed.
Because Indonesia has suffered from a large number of tsunami events since ancient times, some local
cultures developed their own early warning systems. In many cases however, this knowledge has been
lost. As part of the Government of Indonesias efforts, the traditional knowledge systems about tsunami
are being revisited.

The Example of Simelue Island

The effectiveness of local wisdom and culturally shared information about SMONG (local word for
tsunami) was found at the Simelue Island, NAD province of Indonesia after the Indian Ocean tsunami
2004. Simelue is small island located between 2o 53,00 Latitude and 95o 4096,35 Longitude,
approximately 130 km southwest of Sumatra Island in the Indian Ocean. Simelue is about 50 km
southeast of the Sumatra Earthquake epicenter in 2004 and was hit by the ensuing tsunami. The tsunami
wave height varied between 2-15 m. It took only about 5-10 minutes after the receding of sea levels until

Government of Indonesia, 1994.

The buoy system is usually deployed in deep water in the open sea.

- 116 -

the first tsunami wave arrived. This was followed by a second wave that reached a height of around 15m9.
All villages in Alafan district on the northeast coast and facing to the source of the earthquake and
tsunami were destroyed. Although the maximum wave height was measured at only 4 m at Labuhan
Bakti village on the south end of the island, the village suffered large losses in the disaster.
At Simelue, the toll was only 7 people dead and 1 missing from a population of 78, 000 people. This is
less than 0.01% of the population, a figure that stands in sharp contrast to the number of casualties in
Aceh10. The difference is thought to lie in the culturally shared local wisdom about tsunamis. Because
the population is better adapted and has more knowledge about the risks it able to re-act accordingly in
the event.
The tradition of SMONG has been passed down for several generations by the Simeulue people.
Smong is a local word refers to a particularly severe tsunami that struck in 190711. The story of Smong
has been shared by the people of these islands for as long as anyone remembers. Applying the wisdom
passed on through local cultural information exchange networks, the local people interpreted the warning
signs (sensing a quake, seeing the sea level drop, etc.), to mean that one should immediately run to the
hills or high ground. Since the information about SMONG has life saving consequences and therefore of
high value, it was incorporated into the culture as part of local wisdom and survived to save most of the
local population in 200412.
Another example of the effectiveness of locally generated and shared wisdom was during the 2007
Bengkulu tsunami caused by earthquake on 12 September 2007. Even though the tsunami height reached
about 3-4 m in Serangai village there was no loss of life. Learning from the 2004 Aceh tunami, the
coastal community ran to the hills or other safe places when they observed the warning signs.
3.8 Risk Assessment.
A hazard map was developed using technical data from the historical data and numerical forecasting of
tsunami. All information was translated to the GIS platform, significantly refining areas which could
easily be inundated and where specific households, in times of emergencies, need to be evacuated
immediately. They now use GIS as an effective tool for upland and urban planning; and on the drawing
board are plans for appropriate coastal zone development.
Tsunami hazard map shows the information necessary for residents to take refuge to high place or
shelters alocated by local government. The tsunami hazard map usually indicates the inundation area,
infrastructure, land use, evacuation route, and location of safety places.
The Ministry of Research and Technology and The Ministry of Energy and Mineral Resources has
developed their own hazard map for Bali, Padang, Aceh and other areas. Also international agencies
provides expertise and assistance in making hazard maps i.e German-Indonesia Tsunami Early Warning
System (GI-TEWS) for Padang, Cilacap and Bali, vulnerability map and risk map. UNESCO IOC does
scientific survey in North Sumatera Island, community awareness on earthquake and tsunami programs in

This was the highest wave according to witnesses.

10

13.000 people died and 37.000 people were lost

11

According to historical tsunami databases (ITDB, 2004), there was a tsunami in this region in
1907.
12
The message was so well interpreted that the local people even ignored whales left by the
receding water.

- 117 -

Aceh, Nias, Padang, Mentawai. Geo Hazard International: scientific research for geological hazard in
West Sumatera. JICA: work with government institution and universities in developing hazard map and
other activities.

Conclusion

Indonesian coastal areas are prone to tsunami hazards that threaten the sustainability of social and
economic development. The mitigation program is essential for the country and it has been in right
direction under the new laws in disaster management and coastal and small island management.

The mitigation activity covers many aspects. It covers intervention of nine intervention areas: i)
legislation and regulation, ii) integrated coastal planning and management, iii) coastal disaster
mitigation plan, iv) coastal community empowerment,

v) coastal habitat rehabilitation,

vi)

retrofitting, vii) risk assessment, viii) development of evacuation route, and ix) tsunami early warning
system.
-

Enhance coordination between different stakeholders has provided leverages in information sharing,
implementing joint scientific research, and community awareness activities

High occurrence of earthquake and tsunamis provide experience to enhance better disaster
management, coordination, building of retrofitted houses, and community preparedness. Moreover
the societies are now more open to mitigation effort and have a higher willingness to understand
about earthquake and tsunamis.

- 118 -