1008 Tsunami Inundation, Modeling of

larger than the water depth; a non-dispersive, long-

Tsunami Inundation, Modeling of wave model; there is no frequency dispersion in this
PATRICK J. LYNETT model.
Texas A&M University, College Station, USA Tsunami inundation The spatial area flooded as a tsu-
nami rushes inland.

Article Outline
Definition of the Subject
Glossary
Definition of the Subject Tsunami inundation is the one of the final stages of
Introduction tsunami evolution, when the wave encroaches upon and
Brief Review of Tsunami Generation floods dry land. It is during this stage that a tsunami takes
and Open Ocean Propagation the vast majority of its victims. Depending on the proper-
Physics of Nearshore Tsunami Evolution ties of the tsunami (e. g. wave height and period) and the
Effects of Bathymetric and Topographical Features beach profile (e. g. beach slope, roughness), the tsunami
on Inundation may approach as a relatively calm, gradual rise of the ocean
Hydrodynamic Modeling of Tsunami Evolution surface or as an extremely turbulent and powerful bore –
Moving Shoreline Algorithms a wall of white water. The characteristics of this approach
Future Directions determine the magnitude and type of damage to coastal
Bibliography infrastructure and, more importantly, the actions required
of coastal residents to find a safe retreat or shelter.
To gage the nearshore impact of tsunami inundation,
Glossary engineers and scientists rely primarily on three different
Beach profile A cross-shore, or normal to the beach, methods: 1) Field survey of past events, 2) Physical exper-
survey of the seafloor and dry ground elevation imentation in a laboratory, and 3) Numerical modeling.
(bathymetry and topography); a series of spatial loca- It is the last of these methods – numerical simulation of
tion and bottom elevation data pairs. tsunami inundation – that will be the focus of this arti-
Bore A steep hydraulic front which transitions between cle. With numerical simulation, it is possible to predict the
areas of different water level. Tsunamis can approach consequence of future tsunamis on existing coastal towns
land as a turbulent, breaking bore if the incident and cities. This information allows for the establishment
tsunami is of sufficiently large height. of optimum evacuation routes, identification of high-risk
Boussinesq equations An approximate equation model, and/or unprepared areas, re-assessment of building codes
used for waves with wave length of at least two times to withstand wave impact, and placement of tsunami shel-
the local water depth; a long-wave-based model, but ters, for example. It is the hope that, through accurate
includes some frequency dispersion prediction of tsunami effects, in conjunction with pol-
Dispersion, amplitude The separation of wave compo- icy makers willing to implement recommended changes
nents due to a wave-height related difference in wave and a strong public education program, communities will
speed; all else being equal, a wave with a large height show resiliency to tsunami impact, with minimal loss of
will travel faster than one with a small height. life and damage to critical infrastructure.
Dispersion, frequency The separation of wave compo-
nents due to a frequency related difference in wave
Introduction
speed; all else being equal, a wave with a longer period
will travel faster than one with a short period. On December 26, 2004, the boundary between the Indo–
Navier–Stokes equations The full equations of fluid mo- Australian and Eurasian plates off the coast of north-
tion, including dissipation through the fluid molecular ern Sumatra ruptured in a great (Mw 9.3) earthquake at
viscosity only. Other models discussed here, namely 00:58:53 universal time (U.T.). Up to 15m of thrust on
the Shallow Water Wave and Boussinesq equations, the plate interface [31] displaced tens of cubic kilometers
are approximations to these equations. of seawater and propagated a tsunami across the Indian
Runup, or runup height The ground elevation (a vertical Ocean. The earthquake was widely felt throughout South
measure) at the furthest point of landward inundation. Asia and was locally destructive in Sumatra and the An-
Shallow water wave equations An approximate equation daman and Nicobar islands, but it was the tsunami that
model, used for waves with wave length many times caused widespread damage to densely populated coastal
 Tsunami Inundation, Modeling of 1009

communities both nearby and thousands of kilometers undersea earthquake along a great fault length of a sub-
away. duction zone must occur. These earthquakes create large
Due to the extensive damage left behind by large vertical motions of the seafloor. This vertical motion of
tsunamis such as the Indian Ocean tsunami, it is diffi- the seafloor pushes the water above it, essentially creat-
cult if not impossible to put together a complete picture ing a small displacement of water above the earthquake.
of the event with field observations alone. Additionally, This displacement of water will immediately try to spread
for some parts of the world that have not seen a tsunami out and reach a gravitational equilibrium, and it does so as
in recent times, there are no field observations on which waves propagating away from the earthquake zone – this
to develop safety procedures and protect residences from is the tsunami.
future tsunamis. It is for these purposes – understanding To represent the tsunami in numerical models, we use
the detail of tsunami inundation and to estimate tsunami an initial condition. Simply put, there is placed some ir-
hazard – that we must rely on modeling of tsunamis. regular ocean surface profile at the instant after the earth-
There are two primary modeling approaches - physical quake, when the numerical simulations will start. Then,
and numerical. The physical, or experimental, approach based on physics – Newton’s Laws written for fluid – the
uses scaled-down models to look at a particular aspect initial condition evolves and transits oceans. As a tsunami
of a phenomenon. While this approach is integral to the travels unhindered across ocean basins, it does so quickly
fundamental understanding of waves, because of the huge and with little noticeable change. In the deep ocean, even
wavelengths of tsunamis, experiments are limited. For ex- the largest tsunamis have heights only near 1 m and cur-
ample, a tsunami might have a wavelength of 100 km in rents of 10 cm/s, and are not likely to be identified by ships
a deep ocean depth of 1 km, with a wave height of 1 m. or surface buoys in the presence of wind waves.
Note that the above values represent approximate order
of magnitudes for a large subduction zone tsunami. Now,
Physics of Nearshore Tsunami Evolution
to scale this down for the laboratory with a wave tank
depth of 1 m – the tank would have to be 100 m long and A tsunami in the deep ocean is long and travels extremely
the created lab-tsunami would have a hardly measurable fast. As the wave reaches shallow water, near the coast-
height of 1 mm. Numerical modeling, while not “real” in line, the tsunami begins the shoaling process. The speed
the sense that modeling is done on a computer chip with at which long wave such as a tsunami moves, or celerity,
approximated equations of motion rather than in the lab- is a function of the local water depth. The less the depth,
oratory, does not suffer from this scaling problem, and the slower the wave moves. A tsunami, with its very long
can generally accommodate any type of arbitrary wave and length, experiences different water depths at any given in-
ocean depth profile. stant as it travels up a slope; the depth at the front of the
Numerical simulations of tsunami propagation have wave, the portion of the tsunami closest to the shoreline,
been greatly improved in the last 30 years. In the United will generally be in the shallowest water and thus is moving
States, several computational models are being used in the the slowest. The back of the tsunami, on the other hand,
National Tsunami Hazard Mitigation Program, sponsored will be in deeper water and will be moving faster than the
by the National Oceanic and Atmospheric Administration front. This leads to a situation where the back part of the
(NOAA), to produce tsunami inundation maps and pre- wave is moving faster than the front, causing the wave-
dict tsunami runup in real-time for the warning system. length to shorten. With a shortening tsunami, the wave
In addition, there are numerous other models used by re- energy is in essence squeezed into a smaller region, forcing
searchers and engineering companies in an attempt to bet- the height to grow. It is for this reason that, despite having
ter understand tsunami impact. In this article, an overview a height of only a meter in the deep ocean, the tsunami el-
of these models, as well as how they are validated and uti- evation over land can easily exceed 10 m. With this great
lized, is provided. increase in wave height comes a more dynamic and com-
plex phenomenon.
Presented in a more technical manner, a tsunami in
Brief Review of Tsunami Generation
the open ocean is generally a linear, non-dispersive wave.
and Open Ocean Propagation
First, what is meant by a non-dispersive tsunami will be
Before introducing the physics behind propagating discussed. Also, the discussion here will be in terms of
a tsunami across oceans and overland, we must first dis- a large earthquake generation tsunami, such as the 2004
cuss how a tsunami is created. For earthquake generated Indian Ocean event. Other impulsive waves, such as land-
mega-tsunamis, such as the Indian Ocean event, a huge slide or asteroid impact generated waves, are more difficult
1010 Tsunami Inundation, Modeling of

to generalize and will be introduced separately at the end ter depths of 100m and less, the wave can undergo a major
of this section. physical transformation. The properties of this transfor-
Any wave condition, whether it is a tsunami or a typ- mation depend heavily on the characteristics of the beach
ical wind wave in the ocean, can be mathematically de- profile and the wave itself. In the simplest inundation case,
scribed as a superposition, or summation, of a series of the beach profile is relatively steep (footnote: here “steep”
separate sine (or cosine) waves, each with independent should be thought of in terms of the tsunami wavelength.
amplitude and speed. For example, with the right choice If the horizontal distance along the slope connecting deep
of individual sine waves, it is possible to construct even the water to the shoreline is small compared to the tsunami
idealized tsunami: a single soliton. If a wave is considered wavelength, the beach would be considered steep) and
a dispersive wave, then the various sine wave components the tsunami wave height is small, then the runup pro-
will have different wave speeds, and the wave will disperse cess closely resembles that of a wave hitting a vertical wall,
as the faster moving components move away from the and the runup height will be approximately twice the off-
slower ones. If a wave is non-dispersive, then all the com- shore tsunami height. In these special cases, a breaking
ponents move at the same speed, and there is no length- bore front would not be expected; in fact horizontal fluid
wise dispersal, or spreading, of the tsunami wave energy. velocities near the shoreline would be very small. Here,
It is for this reason that tsunamis can be devastating across the tsunami inundation would closely resemble that of
such a large spatial region; the tsunami wave energy will a quickly rising tide with only very minor turbulent, dy-
not disperse but will remain in a focused pulse. namic impacts. However, even in these cases, overland
The dispersion described above is generally what sci- flow constrictions and other features can create localized
entists are referring to during a discussion of dispersive vs energetic inundation.
non-dispersive waves. However, it is more precisely called If the beach profile slope is mild, typical of continental
“frequency dispersion” as it is dominantly dependent on margins, and/or the tsunami wave height is large, then the
the period of the component. There is another type of dis- shallow water evolution process becomes highly nonlin-
persion, called “amplitude dispersion.” This second type ear. However, while the nonlinear effect becomes very im-
of dispersion is a function of the nonlinearity of the wave, portant, in the large majority if cases, frequency dispersion
and is usually discussed under the framework of linear is still very small and can be neglected. Nearshore nonlin-
vs nonlinear waves. For tsunamis, the nonlinearity of the ear evolution is characterized by a strong steepening, and
waves is given by the ratio of the tsunami height to the possible breaking, of the wave front with associated large
water depth. When this ratio is small, such as in the open horizontal velocities. In these cases, turbulent dissipation
ocean, the wave is linear; on the other hand, in shallow wa- can play a major role.
ter the ratio is order unity and thus the wave is no longer While it may be intuitive to postulate that wave break-
linear. The linear/nonlinear nomenclature is not an intu- ing dissipation at the tsunami front plays a significant role
itive physical description of the waves, but comes from the in the tsunami inundation, this may not be altogether cor-
equations describing the tsunami motion, described later rect. This breaking dissipation, while extraordinarily in-
in this section. When this nonlinear effect is taken into tense, is fairly localized at the front which, both spatial
account, it is found that the wave speed is no longer just and temporal, often represents only a small fraction of the
a function of the local depth, but of the wave height as well. tsunami. So, for tsunamis such as the 2004 Indian Ocean
More specifically, looking at two components of the same event, the related dissipation likely had only minor im-
period but with different amplitudes, the component with pact on leading-importance quantities such as the max-
the larger amplitude will have a slightly larger wave speed. imum runup and inland (off-beachfront) flow velocities.
Except for the interesting cases of wave fission, discussed However, the properties of breaking are of great impor-
later in this section, the nonlinear effect of amplitude dis- tance to other aspects of tsunami inundation. The max-
persion does not spread tsunami energy with an end result imum forces on beachfront infrastructure, such as ports,
of lessening nearshore impact; in fact it will act to focus terminals, piers, boardwalks, and houses, should include
wave energy at the front, often leading to a powerful break- the bore impact force as well as the drag force associated
ing bore. with the following quasi-steady flow [54,68]. If one was in-
Thus, open ocean propagation of a conventional terested in understanding how bottom sediments are sus-
tsunami is a relatively uncomplicated process which trans- pended, transported, and deposited by a tsunami, the bore
lates wave energy across basins, subject to wave speed turbulence again may play an important role. Thus, under-
changes that are a function of the local depth. As a tsunami standing the dynamics of a breaking tsunami front is not
enters the nearshore region, roughly characterized by wa- of particular importance for near real-time or operational
 Tsunami Inundation, Modeling of 1011

tsunami forecast models. This information is of great use imum energy flux, and thus intensity, hitting the shore-
for engineers and planners, who can utilize it to design line will decrease. Second, by increasing the duration of
tsunami-resistant structures, for example. the time series, and creating many individual crests, en-
A second energy dissipation mechanism, one that does ergy dissipation can play a bigger role. Using simple en-
play a major role in determining maximum runup, is bot- ergy arguments, it can be shown that, comparing a high-
tom friction. On a fundamental level, this dissipation is density, short-period pulse to a low-density, long-period
caused by the flow interaction with the bottom, where bot- train, more energy will be removed through bottom fric-
tom irregularities lead to flow separations and the result- tion and breaking. This increase will be related to the ra-
ing turbulence. All natural bottoms result in some bottom tio of the period of the entire dispersive wave train to the
friction; a smooth, sandy beach may generate only minor period of the pulse. Numerically studies have shown that
dissipation, while a coral reef or a mangrove forest can play for such cases, the individual wave crests are largely dis-
a huge role in reducing tsunami energy [12]. Such features sipated, and runup is dominated by the carrier wave, or
will be discussed in additional detail in the next section. in other terms it becomes a time-dependent, wave setup
Other means of energy dissipation will be largely local, problem [26].
and may include enhanced mixing due to sediment or de- While a topic of current research in the tsunami
bris entrainment, large shallow-flow vortex generation by community, frequency dispersion may occasionally play
headlands or other natural or artificial obstacles and the a non-negligible role in even the long wavelength, sub-
resulting dissipation, and flow through/around buildings duction zone tsunamis. To date, there have been two cat-
and other infrastructure, sometimes termed macro-rough- egories of argument that dispersion is important for these
ness and grouped with bottom friction. tsunamis: 1) short-period energy generated at the source
Up to this point, we have only discussed the “typical” is significant and leads to different patterns of runup
nearshore tsunami evolution which is portrayed as a wave if included (e. g. [21,28]), and 2) shallow-water nonlin-
without frequency dispersion, and may be called a linear ear interactions can generate short-period components
or nonlinear tsunami, depending on a number of physi- which can become decoupled (or un-locked) from the pri-
cal properties. The rest of this section will be devoted to mary wave, and will change the incident tsunami proper-
those situations where the above characterization may no ties [47].
longer be adequate. Looking first to the tsunami source, Thinking of an arbitrary and complex initial free sur-
waves that are generated by underwater landslides, under- face displacement generated by an undersea earthquake,
water explosions, or asteroid impacts will often not be- there does exist the possibility that dispersive wave energy
have as non-dispersive waves in the open ocean [44,72]. can be initially generated here. This irregular wave condi-
These source regions tend to be at least an order of magni- tion can be constructed as a continuous wave energy spec-
tude smaller in spatial extent compared to their subduc- trum, and by definition there will be finite (albeit small)
tion zone counterparts. Physically, this implies that the energy at all frequencies. The obvious question in this case
generated waves will be of shorter wavelength. As a rule is: what length scale characteristics of the initial free sur-
of thumb, if these generated waves have length scales of face displacement, or the preceding earthquake, will lead
less than 10 times the local depth, then it should be an- to a significant measure of dispersive wave energy? To pro-
ticipated that frequency dispersion will play a role [42]. vide an answer to this question, a simple order-of-mag-
Under this constraint, individual component wave speeds nitude scaling argument is presented here; see Hammack
near the dominant period become frequency dependent. and Segur [19], for example, for a mathematically rigorous
Understanding that an impulsively generated wave can attempt at insight. Let us define a characteristic change in
be dispersive has serious implications. Take, for exam- vertical free surface elevation, , and a horizontal length
ple, a hypothetical landslide located in the Atlantic Ocean scale across which this vertical change occurs, L. Reduc-
which generates a dispersive tsunami (e. g. Ward and Day, ing to the simplest case, a regular wave with wave height
2001). As this tsunami travels across the Atlantic, to ei- equal to , then the wavelength would be 2 L. Follow-
ther the USA east coast or the European west coast, fre- ing this analogy, which will hold in a proportional sense
quency dispersion effects will spread the wave energy in for a Fourier series of wave components, for any  mea-
the direction of propagation. This will convert the initial sured along a tsunami initial condition, there exists a wave
short-period pulse into a long train of waves. By spread- component with wave length equal to 2 L. For that com-
ing this energy out, the inundation impact will be greatly ponent to be significant to the tsunami evolution, the local
reduced. First, by taking a high-density energy pulse and vertical change, , must be some non-negligible fraction
stretching it into a longer, lower-density train, the max- of the maximum tsunami height, H. Note that the H dis-
1012 Tsunami Inundation, Modeling of

cussed here is a global property of the entire initial tsunami essary to describe what a nonlinear, phase-locked wave is.
wave condition. For the individual wave component, with To do this, we will examine the acceleration terms of the
length 2 L, to be dispersive, its length should be less than 1D conservation of momentum equation for the velocity
roughly 10 times the local water depth, h0 . Additionally, component u:
the difference between dispersive wave propagation and
@u @u @p @2 u
a non-dispersive propagation is cumulative. For example, Cu D C 2 : (2)
@t @x @x @x
if the full linear wave theory predicts, or a specific wave
component, a wave speed that is 10% less than the long Now assume that there is a single wave component, under
wave speed, then the predicted arrival time difference will which the velocity oscillates as
grow by 0.1 times the period for each wavelength of propa- cos(kx  ! t) D cos  ; (3)
gation. Thus, the impact of dispersion is related to the dis-
tance of propagation, and is proportional to D/ , where D where k is the wavenumber, ! the frequency, and the
is the total distance traveled by the wave, and is the av- speed of the wave is given by !/k. If the wave is nonlinear,
erage wavelength of the wave across D which can be ex- which is to say that the convective acceleration term in the
pected to be proportional to 2 L. Assuming that is ap- above momentum equation is not negligible, the convec-
proximately equal to 2 L, it can be said that in order tive term will include the product of
for frequency dispersion effects to play a role in tsunami
cos   cos  D cos 2 D cos(2kx  2! t) : (4)
evolution,
ˇ ˇ Thus, through this nonlinear term, a new wave compo-
ˇ h0  ˇ D
max ˇˇ ˇ >ı; (1) nent, with twice the wavenumber and frequency (or half
L2 ˇ H
the wavelength and period) has been generated. From lin-
where ı is some minimum threshold for importance. ear wave theory, it is expected that this new wave, with
What this value should be is an open question, although a shorter period, will have a different wave speed than the
it is likely to be near 0.1. From this term, it is clear the im- original, primary wave. However, from the phase func-
pact of dispersion is a function of a number of the proper- tion of this new wave, there is the speed 2!/2k D !/k,
ties of the initial tsunami condition, and should be taken which is the identical speed of the primary wave. Thus this
into consideration when creating tsunami initial condi- new wave is locked to the phase of the original wave. This
tions. For example, use of discontinuous block-type seg- connection can be rather delicate, and any disruptions to
ments (e. g. [30]), with sharp edges (very small L) may the primary wave, such as a varying seafloor, dissipation,
lead to the conclusion that frequency dispersion is impor- or interactions with other free waves in the train or wave
tant, while it could be a direct result of a coarsely approx- pulse, can cause the new waves to become unlocked. When
imated initial tsunami condition. Also note that this exer- this occurs, the now free waves retain their frequency 2!,
cise does not include the effects of radial spreading, which but take a wavenumber as given by linear wave theory.
could very likely be important for small-scale irregulari- Since these freed waves will be of a shorter period than
ties in the initial condition. Wave height decrease by radial the primary wave, they will travel at a slower speed and
spreading is proportional the horizontal curvature of the generally trail the main wave front.
initial condition and to ( /D)n , and decreases faster for Long wave fission is most commonly discussed in the
dispersive waves (n  1) as compared to non-dispersive literature via a solitary wave propagating over an abrupt
waves (n D 0:5) (e. g. [72]). Thus, for this case of signifi- change in depth, such as a step (e. g. [17,24,33,37,45,55]).
cant radial spreading, it would be very difficult for source- In these cases, there is a deep water segment of the seafloor
based dispersion effects to play a meaningful role in the far profile, where a solitary wave initially exists. In this depth,
field. the solitary wave is of permanent form. As the solitary
Under certain conditions, namely a nonlinear tsunami wave passes over the change in depth, into shallower water,
propagating across a wide shallow shelf, a process called the leading wave energy will try to re-discover a balance
fission may occur. Wave fission is a separation process between nonlinearity and dispersion; the solitary wave.
where wave energy, initially part of a primary wave or Since this new solitary wave will be a different shape and
pulse, attains certain properties, such as higher or lower contain a lower level of mass, by conservation there must
phase speed, that allow it to disconnect from the primary be some trailing disturbance to account for the deficient.
wave and propagate as an independent wave. In the con- This trailing disturbance will take the form of a rank-or-
text of nearshore tsunami evolution, there is a standard dered train of solitons. Figure 1 depicts this process. The
mechanism which is the cause of this fission. First, it is nec- solitons in the trailing train, while smaller in height than
 Tsunami Inundation, Modeling of 1013

Tsunami Inundation, Modeling of, Figure 1

Example of experimental data looking at solitary wave fission by propagation onto a shelf from Goring and Raichlen [17]. Note that
the flume layout and measurement location is given up in the upper right. The initial solitary wave undergoes the fission process and
results in three distinct solitary waves

the leading solitary wave, tend to have a similar wave- as solitary waves, or more generally, cnoidal waves. This
length; this has been shown both analytically, numeri- fact provides some guidance as to the wavelength of these
cally, and experimentally. Note, however, that discussion fission waves; they can be approximately calculated via
of fission in this sense is not particularly relevant to “real” solitary wave theory using the tsunami height and depth of
tsunami modeling, where the offshore wave approach- the shelf. For example, on a shelf with depth of 30 m and an
ing the shelf break rarely resembles a solitary wave solu- incident tsunami height of 5 m, fission waves with a wave-
tion [62]. However, the offshore wave does not need to length of approximately 240 m and period of 13 s would be
specifically be a solitary wave for this process to occur. generated. In recent work looking at tsunamis along the
In numerous eyewitness accounts and videos recorded eastern USA coast, where there exists a wide shallow shelf,
of the 2004 Indian Ocean tsunami, there is evidence of the this fission process has been investigated [14]. Figure 2
tsunami approaching the coastline as a series of short pe- gives a few numerical simulation snapshots, and shows
riod (on the order of 1 min and less) breaking fronts, or where the fission occurs, and the eventually impact on the
bores (e. g. [30]). These short period waves may be the waveform. This simulation, run with the dispersive equa-
result of fission processes of a steep tsunami front prop- tions, generated fission waves with lengths in the range of
agating across a wide shelf of shallow depth. Along the 100–200 m, and required a grid size of 5 m to attain nu-
steep front of a very long period wave, nonlinearity will merically convergent results. In this example, the steep fis-
be very important. There will be a large amount of energy sion waves break offshore, and have little impact on the
in high-frequency components with wavelengths similar maximum runup. A conclusion of this fission issue is that,
the horizontal length of the tsunami front (on the order of if one attempts to simulate tsunami propagation with dis-
1 km). As the wave continues to shoal, the high-frequency persive equations, and if the grid is not chosen to be fine
locked waves may eventually become free waves, and will enough to resolve the short fission waves, the justification
take the form of very short waves“riding” the main wave to use the dispersive model is greatly degraded.
pulse. This situation is akin to an undular bore in a mov-
ing reference frame. This process is, in fact, identical to
Effects of Bathymetric and Topographical Features
that described in the above paragraph, it simply takes place
on Inundation
over a much longer distance. The newly freed waves, in the
nonlinear and shallow environment, will attempt to reach It is well established that large-scale coastal features, such
an equilibrium state, where frequency dispersion and non- as small islands, large shoals, canyons, and shelves, can
linearity are balanced. Thus, the fission waves will appear play am important role in tsunami inundation due to con-
1014 Tsunami Inundation, Modeling of

Tsunami Inundation, Modeling of, Figure 2

Example of tsunami fission. Simulation results are from Geist et al. [14] for a landslide-generated tsunami off the east coast of the
USA. The top plot shows the beach profile and six free surface profiles at different times. The lower subplots are zoom-in’s of those six
profiles, with the times given in the individual plot titles. The red marks visible in the lowest plots indicate regions where the wave is
breaking

ventional shallow water effects such as shoaling and re- tsunami did not have the intensity to remove even beach
fraction (e. g. [4,5,16,36,70]). On the other hand, under- umbrellas. Investigating impacts from the same tsunami,
standing of the impact of smaller scale features is just now Borrero et al. [3], discussed how small scale bathymetry
being developed. This work was largely initiated by field variations affected coastal inundation. One of the con-
observations. Synolakis et al. [60], surveying the coast of clusions of this work was that bathymetry features with
Nicaragua for information about the 1992 tsunami in the length scales 50 m and less had leading order impact on
region, noted that the highest levels of damage along a par- the runup. Looking to the recent Indian Ocean tsunami,
ticular stretch of beach were located directly landward a survey team in Sri Lanka inferred from observations that
of a reef opening used for boat traffic. It was postulated reef and dune breaks lead to locally increased tsunami im-
that the reef gap acted as a lower resistance conduit for pact [38,39]. Also in Sri Lanka, Fernando et al. [12] per-
tsunami energy, behaving like a funnel and focusing the formed a more thorough survey along the southeastern
tsunami. Along neighboring beaches with intact reefs, the coastline, and concluded that there was a compelling cor-
 Tsunami Inundation, Modeling of 1015

relation between coral mining and locally severe tsunami corners of a building, for example, the scour potential of
damage. While additional research is needed to quantify that flow increases greatly, and foundation undermining
the effects of small scale features, the observations hint that is a concern (e. g. see Fig. 3). As with any fluid flow past an
defense measures such as seawalls, once thought to be in- obstacle, the backface of the obstacle is characterized by
consequential to tsunami inundation, may provide some a low-pressure wake. Combined with the interior flood-
protection. ing of a building, this low pressure wake may lead to an
Onshore, tsunami propagation is effected by the gen- outward “pull” force on the back wall, causing it to fail by
eral topography (ground slope), ground roughness, and falling away from the center of the building. Such failures
obstacles (e. g. [41,59,61,67]). The composition of the were observed during field surveys of the 2004 event, as
ground, be it sand, grass, mangroves, or pavement, con- shown in Fig. 4. Increasing the topographical complexity,
trols the roughness and the subsequent bottom friction in built coastal environments, structures are located within
damping. To predict tsunami inundation with high con- close enough proximity to each other such that their dis-
fidence, the ground type must be well mapped and the hy- turbances to the flow may interact. This can lead to irreg-
drodynamic interaction with that type must be well under- ular and unexpected loadings, where for example a 2nd
stood. If the tsunami approaches the shoreline as a bore, row building experiences a larger force than beach front
the process of “bore collapse”, or the conversion of po- buildings due to a funneling effect. These types of interac-
tential to kinetic energy, will cause the fluid to rapidly ac- tions are very poorly understood, and require additional
celerate [56,69]. This fast flow equates to high fluid forces research.
on obstacles such as buildings. Tsunami interaction with
these obstacles can lead to a highly variable local flow
pattern (e. g. [10,67]). As the flow accelerates around the

Tsunami Inundation, Modeling of, Figure 4

Example of damage to the backside of a coastal residence. These
images were taken by the International Tsunami Survey Team to
Sri Lanka in January 2005. The top photograph shows the front
side of the structure, facing the ocean; there is damage but the
Tsunami Inundation, Modeling of, Figure 3 main structure is intact. The lower photo shows the backside of
Example of foundation scour. This image was taken by the Inter- the same building, showing the walls blown out, away from the
national Tsunami Survey Team to Sri Lanka in January 2005 center of the structure
1016 Tsunami Inundation, Modeling of

Hydrodynamic Modeling of Tsunami Evolution and the interaction between tsunamis and coastal struc-
tures, which could be either stationary or movable. At
Numerical simulations of tsunami propagation have made present, stationary coastal structures are parametrized as
great progress in the last thirty years. Several tsunami bottom roughness and contribute to frictional effects in
computational models are currently used in the National these 2DH models. Although by adjusting the roughness
Tsunami Hazard Mitigation Program, sponsored by the and friction parameter satisfactory results can be achieved
National Oceanic and Atmospheric Administration, to for maximum runup and delineation of the inundation
produce tsunami inundation and evacuation maps for zone (e. g., [35]), these models cannot provide adequate
the states of Alaska, California, Hawaii, Oregon, and information for wave forces acting on coastal structures.
Washington. The computational models include MOST As a tsunami propagates into the nearshore region, the
(Method Of Splitting Tsunami), developed originally by wave front undergoes a nonlinear transformation while it
researchers at the University of Southern California [66]; steepens through shoaling. It is in this nearshore region
COMCOT (Cornell Multi-grid Coupled Tsunami Model), that dissipative effects can be important. Bottom friction
developed at Cornell University [35]; and TUNAMI-N2, can play a major role in the maximum runup and area of
developed at Tohoku University in Japan [29]. All three inundation (e. g. [67]). In depth-integrated models, bot-
models solve the same depth-integrated and 2D horizon- tom friction is typically approximated through a quadratic
tal (2DH) nonlinear shallow-water (NSW) equations with (drag) friction term, where the friction factor is calcu-
different finite-difference algorithms. There are a number lated often through a Manning’s coefficient or a Darcy–
of other tsunami models as well, including the finite ele- Wiesbach type friction factor (e. g. [25,35]). The validity of
ment model ADCIRC (ADvanced CIRCulation Model For these steady-flow based coefficients has yet to be rigorously
Oceanic, Coastal And Estuarine Waters; e. g., [53]). For validated for use with tsunamis. If the tsunami is large
a given source region condition, existing models can simu- enough, it can break at some offshore depth and approach
late propagation of a tsunami over a long distance with suf- land as a bore – the white wall of water commonly ref-
ficient accuracy, provided that accurate bathymetry data erenced by survivors of the Indian Ocean tsunami. Wave
exist. breaking in traditional NSW tsunami models has not been
The shallow-water equation models commonly lack handled in a satisfactory manner. Numerical dissipation
the capability of simulating dispersive waves, which, how- is commonly used to mimic breaking (e. g. [36]), and thus
ever, could well be the dominating features in landslide- results become grid dependant. In Boussinesq models, this
generated tsunamis and for the fission processes described breaking is still handled in an approximate manner due to
previously. Several high-order depth-integrated wave hy- the fact that the depth-integrated derivation does not allow
drodynamics models (Boussinesq models) are now avail- for an overturning wave; however these breaking schemes
able for simulating nonlinear and weakly dispersive waves, have been validated for a wide range of nearshore condi-
such as COULWAVE (Cornell University Long and In- tions (e. g. [40]).
termediate Wave Modeling Package; [42]) and FUN- Being depth-integrated, NSW and Boussinesq mod-
WAVE [25]. The major difference between the two is els lack the capability of simulating the vertical details
their treatment of moving shoreline boundaries. Lynett et of many coastal effects, such as strong wave breaking/
al. [44] applied COULWAVE to the 1998 PNG tsunami overturning and the interaction between tsunamis and ir-
with the landslide source; the results agreed with field regularly shaped coastal structures. To address this defi-
survey data well. Recently, several finite element models ciency, several 2D and 3D computational models based
have also been developed based on Boussinesq-type equa- on Navier–Stokes equations have been developed, with
tions (e. g., [73]). Boussinesq models require higher spatial varying degrees of success. An example is COBRAS (Cor-
and temporal resolutions, and therefore are more compu- nell Breaking waves and Structures model Lin and Liu
tationally intensive. Moreover, most of model validation 1998a,b, Lin et al. 1999), which is capable of describing the
work was performed for open-ocean or open-coast prob- interactions between breaking waves and structures that
lems. In other words, the models have not been carefully are either surface piercing or submerged [6,22]. COBRAS
tested for wave propagation and oscillations in semi-en- adopted the Volume of Fluid (VOF) method to track free
closed regions – such as a harbor or bay – especially under surface movement along with a Large Eddy Simulation
resonant conditions. (LES) turbulence closure model; several other computa-
Being depth-integrated and horizontally 2D, NSW and tional models using different free surface tracking meth-
Boussinesq models lack the capability of simulating the ods are also in use, such as the micro surface cell technique
details of many coastal effects, such as wave overturning developed by Johnson et al. [23]. This 3D Navier–Stokes
 Tsunami Inundation, Modeling of 1017

Tsunami Inundation, Modeling of, Figure 5

An simulation snapshot taken from a 3D Navier–Stokes solver with an LES turbulence closure. This setup is looking at a bore impact-
ing a column; the wake and vertical splash are clearly visible (Image provided by Philip L.-F. Liu, Cornell University)

equation model has been tested by two tsunami related ex- ments, their hybrid model compares very well with the
periments. The first is 3D landslide experiments [38,39], completely-3D-domain simulation, requiring roughly 1/5
while the second involves measurements of solitary wave of the total 3D CPU time to achieve similar levels of accu-
forces on vertical cylinders. Both experiments were con- racy. Sittanggang et al. [58] presented work on two-way
ducted in the NEES tsunami basin at Oregon State. An ex- coupling of a Boussinesq model and 2D Navier–Stokes
ample of a LES numerical solution of a solitary wave im- model. These results indicate that there is large poten-
pinging on a circular is shown in Fig. 5. tial for hybrid modeling, in terms of more rapid simu-
Due to their high computational costs, full 3D mod- lation as well as the ability to approach a new class of
els would best be used in conjunction with a depth-inte- problems.
grated 2DH model (i. e., NSW or Boussinesq). While the
2DH model provides incident far-field tsunami informa-
tion, the 3D model computes local wave-structure interac- Moving Shoreline Algorithms
tions. The results from 3D models could also provide a bet- In order to simulate the flooding of dry land by a tsunami,
ter parametrization of small-scale features (3D), which a numerical model must be capable of allowing the shore-
could then be embedded in a large-scale 2DH model. One- line to move in time. Here, the shoreline is defined as the
way coupling (e. g. using a NSW-generated time series to spatial location where the solid bottom transitions from
drive a 3D model, but not permitting feedback from the submerged to dry, and is a function of the two horizontal
3D model back into the NSW) is fairly straightforward to spatial coordinates and time. Numerical models generally
construct (e. g. [18]). Two-way coupling, however, is diffi- require some type of special consideration and treatment
cult and requires consistent matching of physics and nu- to accurately include these moving boundaries; the logic
merical schemes across model interfaces. Previous work in and implementation behind this treatment is called a mov-
this area of two-way coupling of hydrodynamic models is ing shoreline, or runup, algorithm.
limited. Fujima et al. [13] two-way coupled a NLSW model For typical tsunami propagation models, it is possi-
with a fully 3D model. While the results appear promising, ble to divide runup algorithms into two main approaches:
the approach used by Fujima et al. requires ad-hoc and un- those on a fixed grid and those on a Lagrangian or trans-
physical boundary conditions at the model matching lo- formed domain. Both approaches have their advantages
cations, in the form of spatial gradients forced to zero, to and disadvantages; currently fixed grid methods are found
ensure numerical stability. Even with these ad-hoc treat- more commonly in operational-level models (e. g. [66]),
1018 Tsunami Inundation, Modeling of

likely due in large part to their conceptual simplicity. A re- Auxiliary shoreline point methods require dynamic re-
view of these two classes of models will be given in this sec- gridding very near the shoreline, such that the last wet
tion, followed by a review of the standard analytical, exper- point is always located immediately at the shoreline. Ob-
imental, and field benchmarks used to validate the runup viously, this method requires a numerical scheme that can
models. For additional information, the reader is directed readily accommodate non-uniform and changing node lo-
to the comprehensive review given in Pedersen [49]. cations. There is some relation to the extrapolation meth-
With a fixed grid method, the spatial locations of the ods discussed above; the moving shoreline point must
numerical grid points or control volumes are determined be assigned some velocity, and it is extrapolated from
at the start of a simulation, and do not change shape or the neighboring wet points. However, it is fundamen-
location throughout the simulation duration. These meth- tally different in that the shoreline point is explicitly in-
ods can be classified into extrapolation, stairstep, auxili- cluded in the fluid domain. Thus, it would be expected
ary shoreline point, and permeable beach techniques. that the governing conservation equations near the shore-
The extrapolation method has its roots in Sielecki and line are more precisely satisfied here, although still de-
Wurtele [57], with extensions by Hibberd and Pere- pendent on the appropriateness of the extrapolation. One
grine [20], Kowalik and Murty [27], and Lynett et al. [43]. such method can be found in Titov and Synolakis [65],
The basic idea behind this method is that the shoreline lo- and has been successfully applied in NSLW equation
cation can be extrapolated using the nearest wet points, models.
such that its position is not required to be locked onto Another fixed grid treatment of moving boundary
a fixed grid point; it can move freely to any location. The- problems is employing a slot or permeable-seabed tech-
oretically, the extrapolation can be of any order; how- nique [63,64]. Conceptually, this method creates porous
ever, from stability constraints a linear extrapolation is slots, or conduits, through the dry beach, such that there is
generally found. Hidden in the extrapolation, the method always some fluid in a “dry” beach cell, although it may ex-
is roughly equivalent to the use of low-order, diffusive di- ist below the dry beach surface. These porous, “dry” nodes
rectional differences taken from the last wet point into use a modified form of the NLSW; it is noted here that
the fluid domain [43]. Additionally, there are no explicit although in concept this approach is modeling a porous
conservation constraints or physical boundary conditions beach, it is not attempting to simulate the groundwater
prescribed at the shoreline, indicating that large local er- flow under a real, sandy beach, for example. The equations
rors may result if the flow in the extrapolated region can- governing the “dry” domain contain a number of empiri-
not be approximately as linear in slope. The extrapola- cal parameters that are tuned to provide reasonable runup
tion approach can be found in both NLSW and Boussi- agreement with benchmark datasets. The advantage of this
nesq models with finite difference, finite volume, and fi- approach is that is allows the entire domain; including the
nite element solution schemes, and has shown to be ac- fluid and “dry” nodes, to be a determined via a somewhat
curate for a wide range of non-breaking, breaking, two consistent set of governing equations, without requiring
horizontal dimension, and irregular topography problems a direct search routine to determine the shoreline location.
(e. g. [8,9,26,44,51]). The method has gained some popularity in wind wave
Stairstep moving shoreline methods, one of the more models (e. g. [25,46]) when a highly accurate estimate of
common approaches found in tsunami models (e. g. [35]), the shoreline location is not the highest priority. However,
reconstruct the naturally continuous beach profile into the approach has been used with some success in tsunami
a series of constant elevation segments connected through studies (e. g. [30]) despite the fact that the empirical coef-
vertical transitions. In essence, across a single cell width, ficients that govern the model accuracy cannot be univer-
the bottom elevation is taken as the average value. A cell sally determined for a wide range of problems [7].
transitions from a dry cell to a wet cell when the water ele- Alternative to fixed grid methods is the Lagrangian ap-
vation in a neighboring cell exceeds the bottom elevation, proach. Here, the fluid domain is descritized into particles,
and transitions from wet to dry when the local total water or columns of fluid in depth-integrated models, that are
depth falls below some small threshold value. These meth- transported following the total fluid derivative. There are
ods are particularly useful in finite volume and C-grid [1] no fixed spatial grid locations; the columns move freely
type approaches (e. g. [32,36]), but can be difficult to im- in space and time and thus these techniques require nu-
plement in centered difference models, particularly high- merical flexibility, in terms of utilizing constantly chang-
order models or those sensitive to fluid discontinuities, ing space and time steps. The Lagrangian approach can
where the “shock” of opening and closing entire cells can be described as both the more physically consistent and
lead to numerical noise. mathematical elegant method of describing shoreline mo-
 Tsunami Inundation, Modeling of 1019

tion. The shoreline “particle” is included in the physical work. If such a modeling capacity existed, engineering de-
formulation just as any other point in the domain (i. e. no sign of coastal structures could be undertaken in a very
extrapolations are necessary), and thus the shoreline po- efficient manner.
sition accuracy will be compromised only by the overar-
ching physical approximation (e. g. long wave approxima-
Bibliography
tion) and the numerical solution scheme (e. g. second-or-
der time integration). The cost for this accuracy is a math- 1. Arakawa A, Lamb VR (1977) Computational design of the ba-
sic dynamical processes of the UCLA general circulation model.
ematical system that can be more difficult and tedious to
In: Chang J (ed) Methods in computational physics. Academic
solve numerically, typically requiring domain transforma- Press, New York, pp 174–267
tions, mappings, and/or re-griddings. Lagrangian methods 2. Birknes J, Pedersen G (2006) A particle finite element method
have been used successfully in finite difference and finite applied to long wave run-up. Int J Numer Methods Fluids
element nonlinear shallow water (NLSW) and Boussinesq 52(3):237–261
3. Borrero JC, Bourgeois J, Harkins G, Synolakis CE (1997) How
equation models (e. g., [2,15,48,50,52,74]).
small-scale bathymetry affected coastal inundation in the 1992
Nicaraguan tsunami. Fall AGU Meeting, San Francisco
4. Briggs MJ, Synolakis CE, Harkins GS, Green D (1994) Laboratory
Future Directions
experiments of tsunami runup on a circular island. PAGEOPH
Towards a more robust simulation of tsunami inundation, 144(3/4):569–593
there are two major issues which require additional fun- 5. Carrier GF (1966) Gravity waves on water of variable depth.
Fluid J Mech 24:641–659
damental investigations: dissipation mechanisms and in- 6. Chang K-A, Hsu T-J, Liu PL-F (2001) Vortex generation and evo-
teraction with infrastructure. Bottom friction, known to lution in water waves propagating over a submerged rectan-
play an important role in inundation, needs to be re-ex- gular obstacle. Part I solitary waves. Coast Eng 44:13–36
amined starting from its basic formulation. Can a steady- 7. Chen Q, Kirby JT, Dalrymple RA, Kennedy AB, Chawla A (2000)
flow based Mannings-type expression for bottom friction Boussinesq modeling of wave transformation, breaking, and
runup: Part I 2D. J Waterw Port Coast Ocean Eng 126(1):
be used for tsunami? Does the unsteady nature of the 57–62
tsunami flow make use of these existing formulations in- 8. Cheung K, Phadke A, Wei Y, Rojas R, Douyere Y, Martino C,
valid? The answer to these questions may be different de- Houston S, Liu PL-F, Lynett P, Dodd N, Liao S, Nakazaki E (2003)
pending on what part of the wave is investigated (e. g. Modeling of storm-induced coastal flooding for emergency
front). In addition, the hydrodynamic effect of common management. Ocean Eng 30:1353–1386
9. Cienfuegos R, Barthelemy E, Bonneton P (2007) A fourth-or-
coastal vegetation, such as mangroves, needs to be quan- der compact finite volume scheme for fully nonlinear and
tified. There is current discussion of the use of such nat- weakly dispersive Boussinesq-type equations. Part II: Bound-
ural roughness as a tsunami defense (e. g. [11]); confi- ary conditions and model validation. Int J Numer Meth Fluids
dence cannot be put in such measures until it is under- 53(9):1423–1455
10. Cross RH (1967) Tsunami surge forces, ASCE Jl Waterways &
stood how they behave. In addition to bottom friction,
Harbors Division WW4:201–231
which exists at all locations and times under an inundat- 11. Danielsen F, Sørensen MK, Olwig MF, Selvam V et al (2005) The
ing tsunami, wave breaking can increase the total energy Asian tsunami: A protective role for coastal vegetation. Science
dissipation. While breaking is generally confined to the 310:643
leading front of a tsunami, the characteristics of this front 12. Fernando HJS, McCulley JL, Mendis SG, Perera K (2005) Coral
poaching worsens tsunami destruction in Sri Lanka. Eos Trans
are important for hydrodynamic loadings on beachfront
AGU 86(33):301
structures, and may be significant to the net sediment and 13. Fujima K, Masamura K, Goto C (2002) Development of the
debris transport of a tsunami. Three-dimensional tsunami 2d/3d hybrid model for tsunami numerical simulation. Coastal
breaking is poorly understood and has received little atten- Eng J 44(4):373–397
tion. 14. Geist E, Lynett P, Chaytor J (2008) Hydrodynamic modeling
Wave loadings and interactions with infrastructure are of tsunamis from the currituck landslide, Marine Geology (in
press)
not well understood. To tackle this problem, tsunami hy- 15. Gopalakrishnan TC, Tung CC (1983) Numerical analysis of
drodynamic models need to be coupled with structural a moving boundary problem in coastal hydrodynamics. Intl J
and geotechnical models. Ideally, these models should all Numer Meth Fluids 3:179–200
be two-way coupled, such that the displacement of a struc- 16. González FI, Satake K, Boss EF, Mofjeld HO (1995) Edge
ture, be it a single collapsed wall, will change the flow wave and non-trapped modes of the 25 April 1992 Cape
Mendocino tsunami. Pure Appl Geophys PAGEOPH 144(3–4):
pattern, and scour underneath the foundation will change 409–426
the structure stability. Additionally, impacts of flow-trans- 17. Goring DG, Raichlen F (1992) Propagation of long waves onto
ported debris (e. g. cars) should be included in this frame- shelf. J Waterw Port Coast Ocean Eng 118(1):43–61
1020 Tsunami Inundation, Modeling of

18. Guignard S, Grilli ST, Marcer R, Rey V (1999) Computation of 36. Liu PL-F, Cho Y-S, Briggs MJ, Kanoglu U, Synolakis CE (1995)
shoaling and breaking waves in nearshore areas by the cou- Runup of solitary waves on a circular island. J Fluid Mech
pling of BEM and VOF methods. Proc. 9th Offshore and Polar 320:259–285
Engng. Conf., vol III, ISOPE99, Brest, France, pp 304–309 37. Liu PL-F, Cheng Y (2001) A numerical study of the evolution of
19. Hammack J, Segur H (1978) Modelling criteria for long water a solitary wave over a shelf. Phys Fluids 13(6):1660–166
waves. J Fluid Mech 84(2):359–373 38. Liu PL-F, Lynett P, Fernando H, Jaffe BE, Fritz H, Higman B,
20. Hibberd S, Peregrine DH (1979) Surf and run-up on a beach. Morton R, Goff J, Synolakis C (2005) Observations by the In-
J Fluid Mech 95:323–345 ternational Tsunami Survey Team in Sri Lanka. Science 308:
21. Horrillo, Juan, Kowalik, Zygmunt, Shigihara, Yoshinori (2006) 1595
Wave dispersion study in the Indian Ocean–Tsunami of De- 39. Liu PL-F, Wu T-R, Raichlen F, Synolakis CE, Borrero JC (2005)
cember 26, 2004. Marine Geodesy 29(3):149–166(18) Runup and rundown generated by three-dimensional sliding
22. Hsu T-J, Sakakiyama T, Liu PL-F (2002) A numerical model for masses. J Fluid Mech 536:107–144
waves and turbulence flow in front of a composite breakwater. 40. Lynett P (2006) Nearshore modeling using high-order Boussi-
Coast Eng 46:25–50 nesq equations. J Waterway Port Coastal Ocean Eng (ASCE)
23. Johnson DB, Raad PE, Chen S (1994) Simulation of impacts of 132(5):348–357
fluid free surfaces with solid boundaries. Int J Num Methods 41. Lynett P (2007) The effect of a shallow water obstruction on
Fluids 19:153–176 long wave runup and overland flow velocity. J Waterway Port
24. Johson RS (1972) Some numerical solutions of a variable-co- Coastal Ocean Eng (ASCE) 133(6):455–462
efficient Korteweg–de Vries equation (with application to soli- 42. Lynett P, Liu PL-F (2002) A numerical study of submarine
tary wave development on a shelf). J Fluid Mech 54:81 landslide generated waves and runup. Proc R Soc London A
458:2885–2910
25. Kennedy AB, Chen Q, Kirby JT, Dalrymple RA (2000) Boussinesq
43. Lynett P, Wu T-R, Liu PL-F (2002) Modeling wave runup with
modeling of wave transformation, breaking, and runup. 1: 1D.
depth-integrated equations. Coast Eng 46(2):89–107
J Waterway Port Coastal Ocean Eng ASCE 126:39–47
44. Lynett P, Borrero J, Liu PL-F, Synolakis CE (2003) Field survey
26. Korycansky DG, Lynett P (2007) Runup from impact tsunami.
and numerical simulations: A review of the 1998 Papua New
Geophys J Int 170:1076–1088
Guinea tsunami. Pure Appl Geophys 160:2119–2146
27. Kowalik Z, Murty TS (1993) Numerical simulation of two-di-
45. Madsen OS, Mei CC (1969) The transformation of a solitary
mensional tsunami runup. Marine Geodesy 16:87–100
wave over an uneven bottom. J Fluid Mech 39:781
28. Kulikov E (2005) Dispersion of the Sumatra tsunami waves in
46. Madsen PA, Sorensen OR, Schaffer HA (1997) Surf zone dynam-
the Indian Ocean detected by satellite altimetry. Report from
ics simulated by a Boussinesq-type model: Part I. Model de-
P.P. Shirshov Institute of Oceanology, Russian Academy of Sci-
scription and cross-shore motion of regular waves. Coast Eng
ences, Moscow
32:255–287
29. Imamura F (1995) Review of tsunami simulation with a finite 47. Matsuyama M, Ikeno M, Sakakiyama T, Takeda T (2007) A study
difference method, long-wave runup models. World Scien- on tsunami wave fission in an undistorted experiment. Pure
tific, Singapore, pp 25–42, http://en.wikipedia.org/wiki/2004_ Appl Geophys 164:617–631
Indian_Ocean_earthquake 48. Özkan-Haller HT, Kirby JTA (1997) Fourier-Chebyshev colloca-
30. Ioualalen M, Asavanant JA, Kaewbanjak N, Grilli ST, Kirby JT, tion method for the shallow water equations including shore-
Watts P (2007) Modeling of the 26th December 2004 Indian line run-up. Appl Ocean Res 19:21–34
Ocean tsunami: Case study of impact in Thailand. J Geophys 49. Pedersen G (2006) On long wave runup models. In: Proceed-
Res 112:C07024, doi:10.1029/2006JC003850 ings of the 3rd International Workshop on Long-Wave Runup
31. Lay T, Kanamori H, Ammon C, Nettles M, Ward S, Aster R, Beck Models, in Catalina Island, California
S, Bilek S, Brudzinski M, Butler R, DeShon H, Ekström G, Satake 50. Pedersen G, Gjevik B (1983) Runup of solitary waves. J Fluid
K, Sipkin S (2005) The great Sumatra–Andaman earthquake Mech 142:283–299
of December 26, 2004. Science 308:1127–1133, doi:10.1126/ 51. Pedrozo-Acuna A, Simmonds DJ, Otta AK, Chadwick AJ (2006)
science.1112250 On the cross-shore profile change of gravel beaches. Coast Eng
32. LeVeque RJ, George DL (2004) High-resolution finite volume 53(4):335–347
methods for the shallow water equations with bathymetry and 52. Petera J, Nassehi V (1996) A new two-dimensional finite el-
dry states. In: Liu PL, Yeh H, Synolakis C (eds) Advanced numer- ement model for the shallow water equations using a La-
ical models for simulating tsunami waves and runup. vol 10 of grangian framework constructed along fluid particle trajecto-
Advances in Coastal and Ocean Engineering. World Scientific, ries. Int J Num Methods Eng 39:4159–4182
Singapore 53. Priest GR et al (1997) Cascadia subduction zone tsunamis: Haz-
33. Losada MA, Vidal V, Medina R (1989) Experimental study of the ard mapping at Yaquina Bay, Oregon. Final Technical Report to
evolution of a solitary wave at an abrupt junction. J Geophys the National Earthquake Hazard Reduction Program DOGAMI
Res 94:14557 Open File Report 0-97-34, p 143
34. Liu PL-F, Synolakis CE, Yeh H (1991) Report on the interna- 54. Ramsden JD (1996) Forces on a vertical wall due to long waves,
tional workshop on long-wave run-up. J Fluid Mech 229:675– bores, and dry-bed surges. J Waterway Port Coast Ocean Eng
688 122(3):134–141
35. Liu PL-F, Cho Y-S, Yoon SB, Seo SN (1994) Numerical simula- 55. Seabra-Santos FJ, Renouard DP, Temperville AM (1987) Nu-
tions of the 1960 Chilean tsunami propagation and inunda- merical and experimental study of the transformation of a soli-
tion at Hilo, Hawaii. in: El-Sabh MI (ed) Recent development in tary wave over a shelf or isolated obstacles. J Fluid Mech
tsunami research. Kluwer, Dordrecht, pp 99–115 176:117
 Tsunami Inundation, Modeling of 1021

56. Shen M, Meyer R (1963) Climb of a bore on a beach Part 3. Run- 66. Titov VV, Synolakis CE (1998) Numerical modeling of tidal wave
up. J Fluid Mech 16(1):113–125 runup. J Waterway Port Coast Ocean Eng ASCE 124(4):157–
57. Sielecki A, Wurtele MG (1970) The numerical integration of the 171
nonlinear shallow-water equations with sloping boundaries. 67. Tomita T, Honda K (2007) Tsunami estimation including effect
J Comput Phys 6:219–236 of coastal structures and buildings by 3D model. Coastal Struc-
58. Sittanggang K, Lynett P, Liu P (2006) Development of a Boussi- tures ’07, Venice
nesq-RANSVOF Hybrid Wave Model. in Proceedings of 30th 68. Yeh H (2006) Maximum fluid forces in the tsunami runup zone.
ICCE, San Diego, pp 24–25 J Waterway Port Coast Ocean Eng 132(6):496–500
59. Synolakis CE (1987) The runup of solitary waves. J Fluid Mech 69. Yeh H, Ghazali A, Marton I (1989) Experimental study of bore
185:523–545 runup. J Fluid Mech 206:563–578
60. Synolakis CE, Imamura F, Tsuji Y, Matsutomi S, Tinti B, Cook B, 70. Yeh H, Liu P, Briggs M, Synolakis C (1994) Propagation and
Ushman M (1995) Damage, Conditions of East Java tsunami amplification of tsunamis at coastal boundaries. Nature 372:
of 1994 analyzed. EOS, Transactions, American Geophysical 353–355
Union 76(26):257 and 261–262 71. Ward SN, Day S (2001) Cumbre Vieja Volcano – Potential col-
61. Tadepalli S, Synolakis CE (1994) The runup of N-Waves on slop- lapse and tsunami at La Palma, Canary Islands. Geophys Res
ing beaches, Proc R Soc London A 445:99–112 Lett 28(17):3397–3400
62. Tadepalli S, Synolakis CE (1996) Model for the leading waves of 72. Weiss R, Wunnemann K, Bahlburg H (2006) Numerical mod-
tsunamis. Phys Rev Lett 77:2141–2144 elling of generation, propagation, and run-up of tsunamis
63. Tao J (1983) Computation of wave runup and wave breaking. caused by ocean impacts: model strategy and technical solu-
Internal Report. Danish Hydraulics Institute, Denmark tions. Geophys J Int 67:77–88
64. Tao J (1984) Numerical modeling of wave runup and breaking 73. Woo S-B, Liu PL-F (2004) A finite element model for modified
on the beach. Acta Oceanol Sin 6(5):692–700 (in chinese) Boussinesq equations. Part I: Model development. J Waterway
65. Titov VV, Synolakis CE (1995) Modeling of breaking and non- Port Coast Ocean Eng 130(1):1–16
breaking long wave evolution and runup using VTCS-2. J Har- 74. Zelt JA (1991) The run-up of nonbreaking and breaking solitary
bors Waterways Port Coast Ocean Eng 121(6):308–316 waves. Coast Eng 15(3):205–246