Sugawara 2014

Marine Geology 352 (2014) 295–320
Contents lists available at ScienceDirect
Marine Geology
journal homepage: www.elsevier.com/locate/margeo
Numerical models of tsunami sediment transport — Current

understanding and future directions
Daisuke Sugawara a, Kazuhisa Goto a,⁎, Bruce E. Jaffe b
a
International Research Institute of Disaster Science, Tohoku University, 980-8579, Japan
b
US Geological Survey, Santa Cruz, CA, 95060, USA
a r t i c l e i n f o a b s t r a c t
Article history: Researchers who study tsunami deposits share common ultimate goals of their work, which are to better assess
Received 9 June 2013 the magnitude information of paleotsunamis and to contribute to the assessment of future tsunami risks. Numer-
Received in revised form 31 January 2014 ical modeling of tsunami sediment transport is an important piece of interdisciplinary research that fills the gap
Accepted 15 February 2014
between geological studies and practical utilization of tsunami deposits. Forward and inverse numerical models
Available online 26 February 2014
that address tsunami transport of sand and boulders have been developed over the last two decades. Forward
Keywords:
models are capable of delineating the time evolution of tsunami hydrodynamics, sediment transport and the
Numerical modeling resulting morphological changes associated with erosion and deposition. Inverse models estimate tsunami char-
Tsunami deposit acteristics, such as flow speed and depth, from deposits. Numerical modeling can be used not only to quantify
Sediment transport paleotsunamis but also to enhance our understanding of tsunami sedimentology and hydrodynamics. To make
Boulder progress towards the ultimate goal of improved tsunami risk assessment, development of an in-depth mutual
understanding between modelers and geologists of the advantages, limitations and uncertainties in both numer-
ical modeling and geological records is an important challenge.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction considered among the diagnostic characteristics. The sand layer typical-
ly has a sharp, erosive bottom contact with the underlying sediment and
Tsunamis possess an enormous amount of hydrodynamic energy may include the eroded sediments as rip-up clasts. Marine microfossils,
that is capable of transporting fine- to coarse-grained clastic materials, such as diatoms and foraminifera, are frequently included into the tsu-
including meter-long boulders, distances of several hundreds of meters nami deposits indicating a marine source for the sediments. Vertical
to as far as a few kilometers inland. These materials can be preserved as variation in grain-size of layers often exhibit normal grading, which
tsunami deposits. Since the first geological documentation of tsunami can be interpreted as the result of deposition from suspension (Jaffe
sedimentation in the 1950s and 1960s (e.g., Shepard et al., 1950; et al., 2012). Of course the nature of tsunami deposits can be both
Kon'no, 1961; Coleman, 1968; Makino, 1968), many examples of mod- more complex or even simpler, reflecting local tsunami hydrodynamics
ern and paleotsunami deposits have been reported from coastal areas and geological context. The diversity of composition, spatial distribution
worldwide (e.g. Dawson and Shi, 2000; Dawson and Stewart, 2007; and sedimentary structure of tsunami deposits are now widely recog-
Goff et al., 2012). Among the various grain sizes that form tsunami de- nized, and together with dating and correlation for paleotsunami de-
posits, sandy tsunami deposits have been intensively studied because posits, inform a multi-proxy approach that improves the ability to
they are common and sand layers can be recognized easily within assign a tsunami origin to deposits (Goff et al., 2012). The multi-proxy
peaty marsh deposits in coastal lowlands, the setting of many studies. approach requires a comprehensive and interdisciplinary investigation
Observation of boulder deposition by recent tsunami events has stimu- through the collaboration of specialists from different research fields,
lated research on the identification and characterization of these de- such as geology, sedimentology, geomorphology, biology, geochemistry
posits (e.g. Goto et al., 2007). Muddy and gravelly deposits have been and archeology (Goff et al., 2012).
less studied, but they are important for understanding the history and The ultimate goal of tsunami deposit research is to be better able to
inundation area of paleotsunami events (e.g. Chagué-Goff et al., 2011). assess tsunami hazards (Jaffe et al., 2012). The assessment quantifies the
Post-tsunami geological surveys have established common charac- magnitude information of the tsunami, including the inundation area,
teristics of tsunami deposits. For example, a general landward thinning run-up height and flow depth and speed. This information can be uti-
and fining is frequently reported for sandy tsunami deposits and are lized for disaster mitigation activities, such as designing tsunami protec-
tive structures, developing hazard maps, and planning evacuation areas
⁎ Corresponding author. and routes. Therefore, quantification of a tsunami event based on its de-
E-mail address: goto@irides.tohoku.ac.jp (K. Goto). posits has been an active research area since the recognition of the
http://dx.doi.org/10.1016/j.margeo.2014.02.007
0025-3227/© 2014 Elsevier B.V. All rights reserved.
296 D. Sugawara et al. / Marine Geology 352 (2014) 295–320
ubiquitous distribution and usefulness of paleotsunami deposits (e.g. deposits (Cheng and Weiss, 2013). Other studies explore the potential
Bourgeois and Minoura, 1997; Huntington et al., 2007; Bourgeois, of numerical modeling for assessing tsunami hydrodynamics (e.g. Jaffe
2009). Earthquakes generate up to 76% of historical tsunamis (National and Gelfenbaum, 2007), as well as to improve understanding of the pro-
Geophysical Data Center / (NGDC/WDS) Global Historical Tsunami Da- cesses and controlling factors for tsunami sediment transport (e.g.
tabase; Available at http://www.ngdc.noaa.gov/hazard/tsu_db.shtml). Apotsos et al., 2011b; Gusman et al., 2012). In this context, the nu-
Determining the magnitude of past earthquakes is one of essential merical modeling of tsunami sediment transport is an expanding
parts for the long-term forecasting of future earthquakes, which is in area of research because of its potential to answer many questions
particular important for coastal areas facing subduction zones (Satake related to the quantification of tsunami events (e.g. Peters et al.,
and Atwater, 2007). A number of historical subduction zone earth- 2007; Tappin, 2007; Bourgeois, 2009; Goff et al., 2012).
quakes are known to have triggered large-scale tsunamis that left sedi- In contrast to geological research, the numerical modeling of tsuna-
mentary deposits, and researchers have studied paleotsunami deposits mi sediment transport is a relatively new field that started in the 1990s
in those regions to estimate the magnitude of paleoseismic events (mostly after the late 2000s), despite its importance (Huntington et al.,
(e.g. Bourgeois and Minoura, 1997). The significance of studying tsuna- 2007). This is because the modeling of tsunami sediment transport re-
mi deposits has been increasing over the last two decades, particularly quires an integration of advanced knowledge on tsunami geology, hy-
after the Mw 9.3 Sumatra Earthquake and subsequent Indian Ocean tsu- drodynamics and hydraulics, which is ideally achieved through the
nami (IOT) in 2004, which caused massive devastation and loss of life to collaboration of geologists, oceanographers and coastal engineers. This
the coastal areas surrounding the Indian Ocean. Despite progresses in paper is intended to review the recent progress of the numerical model-
research that utilizes the geologic record to assess tsunami hazard, we ing of tsunami sediment transport for sandy and boulder deposits. Un-
still have been experiencing many subsequent tsunami disasters, in- fortunately, there have been no numerical investigations of tsunami-
cluding the 2009 Samoa, the 2010 Chile and the 2011 Tohoku-oki induced transport of mud and gravels. The current understanding and
tsunamis. related problems are summarized, and future directions are suggested
There are outstanding questions about how to utilize tsunami de- to encourage the application of the numerical modeling for future tsu-
posits to quantify the magnitude of a tsunami event. For example, nami risk assessment.
does the distribution of sandy deposit reflect the tsunami hydrodynam-
ics, and if so, is it possible to determine the tsunami inundation area or
run-up from the deposit distribution? MacInness et al. (2009) summa- 2. Brief overview of the tsunami sediment transport models
rized sedimentary data collected during post-tsunami field observations
at low-relief coastlines and showed that the tsunami deposits common- Two different modeling approaches, forward and inverse, are used
ly extend to 90% of water run-up and inundation limits. They argued to investigate tsunami sediment transport (Fig. 1). A forward model
that the maximum inland extent of sandy deposits can be used as a consists of a coupled system of tsunami source, hydrodynamic and sed-
proxy for tsunami run-up and inundation, given that the pre-tsunami iment transport models (Fig. 1a). Note that, in order to run any kind
shoreline position can be reconstructed. The finding has been used for of forward models, we need either a realistic or idealized bathymetry/
the hydrodynamic modeling of some modern tsunamis to constrain topography model, which comprise one or more computational do-
the geometry and focal parameters of the causal earthquake events mains. For the case of modeling of an earthquake tsunami when focal
(Martin et al., 2008; MacInnes et al., 2010). Studies on modern tsunamis parameters are specified, the initial tsunami waveform can be calculat-
in other areas also demonstrated that the maximum elevation reached ed using analytical solutions from elastic models that calculate crustal
by a tsunami and the maximum extent of inundation is often beyond deformation due to the fault motion (e.g. Mansinha and Smylie, 1971;
the maximum extent of tsunami deposits, and tsunami deposits may Okada, 1985). Alternatively, external forcing, such as idealized or ob-
be discontinuous near the limit of inundation (Peters et al., 2007). Re- served tsunami waveforms, is directly introduced to the computational
cent studies have found that the maximum inland extent of sandy tsu- domain. The time evolution of tsunami propagation and inundation is
nami layers sometimes is only 50–60% of the inundation distance then simulated by the tsunami hydrodynamic model, which is com-
(Goto et al., 2011a; Abe et al., 2012). Absence of confidence in the corre- posed of an equation of continuity and equations of momentum conser-
spondence between inundation and deposit distribution may raise a vation. The nonlinear shallow-water equations (NLSWE) are commonly
considerable uncertainty in the assessment of the tsunami hazard. To used as the equation of motion, under an assumption of hydrostatic ap-
develop a basis for determining the tsunami inundation and run-up proximation. In the two-dimensional horizontal (2DH) models, the hor-
from the distribution of sandy deposits, for example, a better under- izontal motion of the water is assumed to be uniform throughout the
standing on the process and controlling factors of tsunami-induced sed-
iment transport and deposition is needed.
The numerical modeling of tsunami sediment transport is an impor-
tant piece of interdisciplinary research that is attempting to improve
understanding of tsunami hydrodynamics and sedimentology as well
as to fill the gap between the geological studies and practical utilization
of tsunami deposits (Huntington et al., 2007; Weiss and Bourgeois,
2012). Numerical modeling explicitly quantifies tsunami magnitude in-
formation, depicts the physical processes of sediment transport and is
able to predict the resulting erosion and deposition. With regard to
the gap between the extent of sandy deposits and water inundation,
for example, Cheng and Weiss (2013) carried out a series of numerical
experiment using a tsunami hydrodynamic model coupled with a sim-
plified sediment deposition model. They found that the initial amplitude
(i.e., offshore wave amplitude) and onshore slope influence the differ-
ence between maximum inundation and sand extent more than the
grain size of the sand. They illustrated the expected ratio of sediment
extent to seawater inundation as a function of the initial wave ampli-
tude and beach slope. This concept may be utilized to derive a first- Fig. 1. General framework of forward and inverse numerical models of tsunami sediment
order estimate of the maximum inundation of a tsunami from its transport.
D. Sugawara et al. / Marine Geology 352 (2014) 295–320 297
water column, and flow speed is calculated as a depth-averaged value. If
Ashida and Michiue (1972), Takahashi et al. (2000)

the problem is simple, the 2DH model can be reduced to a one-
Takahashi et al. (2000), Takahashi et al. (2008),

dimensional vertical (1DV) model. The advantage of a 2DH/1DV
Takahashi et al. (2000), Yoshii et al. (2010)
Yoshii et al. (2010), Gusman et al. (2012)

model is the relatively small computational load and seamless calcula-
tion of tsunami generation in the offshore, propagation to the nearshore,
and inundation on the land. However, because sediment transport under
Kesen-numa, Lhok Nga, Kirinda

a wave/flow field is a function of the flow speed near the bottom, 2DH/
1DV models have limitations for application to tsunami sediment trans-
TUNAMI-N2 and others
Sand, single grain-size

port. Three-dimensional (3D) models have been developed to resolve
the vertical profile of flow speed and complex wave/flow field. The verti-
Dietrich (1982)
cal profile of the flow can be resolved also by a two-dimensional vertical
(2DV) model. The wave/flow field calculated by the hydrodynamic
model is applied to the sediment transport model, and the temporal
2DH
STM
progress of erosion and deposition of sediment are computed. Sedi-
ment transport causes the changes in the bathymetry and topogra-
phy and/or the movement of boulders. The morphological change
Kihara and Matsuyama (2010)

simulated by the modeling of a sandy tsunami deposit is fed back
into the hydrodynamic model.

Inverse models do not require specification of tsunami source or use
van Rijn (1984b)

van Rijn (1984a)
3D (hydrostatic)
hydrodynamic models (Fig. 1b). Tsunami deposit characteristics are
Soulsby (1997)
input to the model, and tsunami characteristics, such as flow speed
C-HYDRO3D
C-HYDRO3D
and height, are the output. Inverse models employ either analytical or
Kirinda
numerical approaches. For analytical inverse models, some assumptions
and/or simplifications of the tsunami hydrodynamics are introduced
to derive tsunami height and/or flow speed (e.g., Moore et al.,
Ontowirjo et al. (2012)

Ontowirjo et al. (2012)

2007; Soulsby et al., 2007). For numerical approaches, the model
Ribberink (1998)
internally calculates the transport and deposition of sediment under a
van Rijn (1984b)

van Rijn (1984b)
given flow field, and iteratively fits simulated deposits to observed
deposits to calculate tsunami characteristics (e.g., Jaffe and Gelfenbaum,
Lhok Nga
COMCOT
2007).
2DH
3. Forward modeling of sandy tsunami deposits
van Rijn (1993), Soulsby (1997)

3.1. Background and general framework of the forward models
Sand, mixed grain-size

A purpose of forward numerical modeling of tsunami hydrodynam-
Li et al. (2012a,b)
ics is to understand how tsunamis behave and interact with bathymetry
van Rijn (1993)

and topography. By doing so, we are better able to predict where and to
Lhok Nga
what degree the wave and flow will impact coastal areas. Developing
COMCOT
XBeach
real-time tsunami forecasting systems is a good example of an efficient

2DH
practical application of forward modeling (e.g., Titov et al., 2005; Wei

et al., 2008). Since the early stages of this research, hydrodynamic
Gelfenbaum et al. (2007), Apotsos et al. (2011a,b,c)
modeling has investigated the heights of historical tsunamis

(e.g., Aida, 1977). Important insights can result from forward modeling
of propagation and inundation of paleotsunami events, such as the
van Rijn (1993), van Rijn et al. (2004)
1700 Cascadia (Satake et al., 2003) and the AD 869 Jogan tsunamis
(e.g., Satake et al., 2008; Namegaya et al., 2010; Sugawara et al., 2011).
Summary of available numerical models of tsunami sediment transport.
The development of forward numerical models for the transport of

sandy sediment is driven by the value of improving understanding of
Fagafue Bay, Kuala Meurisi
sediment-related effects for modern tsunamis and advancing hazard as-

Sand, mixed grain-size
2DV/3D (hydrostatic)
sessment. Early modeling investigated the impacts of tsunami sediment

van Rijn (2007a,b)
van Rijn (1993)
transport in coastal areas, such as erosion and deposition of ports and

coastal farmlands, and formation of modern and paleotsunami deposits
(Takahashi et al., 1993, 2000). Recent publications have emphasized the
Delft3D
Delft3D
significance and potential of forward modeling to explore paleotsunami

deposits in inner bays (Kihara et al., 2012; Li et al., 2012a) and to assess
tsunami hazards (Apotsos et al., 2011a; Gusman et al., 2012; Li et al.,
Formulation for suspended load
2012b; Ontowirjo et al., 2012; Ranasinghe et al., 2013). To date, research

Formulation for settling rate
Sediment transport model
on forward modeling mostly focuses on deposition and processes of

Formulation for bed load
sediment transport in recent tsunamis.

Hydrodynamic model
Table 1 summarizes the existing forward numerical models that

have been applied to the transport of sandy sediment by tsunamis.
Particle size
These previous studies include investigations of three modern tsunami

Dimension
Case study
Reference
events (the 1960 Chile, the 2004 Indian Ocean and the 2009 Samoa
Table 1
tsunamis), as well as flume experiments (idealized topography) and a

case study on projected earthquake-tsunami scenarios.
For sediment transport modeling of sandy tsunami deposits, sediment 3.2.1.1. Where and when do tsunamis erode and deposit sediment? The in-
is treated as continuum. Erosion (pick-up) and deposition (settling and teraction between tsunami hydrodynamics and sediment transport has
transport convergence) of sediment is computed as changes in the been a primary focus of tsunami deposit research (e.g., Shiki et al.,
mass, volume or concentration of sediment occupying a unit area or 2008). One of the advantages of forward modeling is that it is capable
volume. The mode of sediment transport is classified into two different of delineating the time evolution of sediment transport, which is useful
categories: (1) bedload, and (2) suspended load (e.g., Allen, 1984). for answering questions raised by sedimentological investigations of
Bedload is transported in rapid interactions with the bottom surface, tsunami deposits. For example, Li et al. (2012b) illustrated the inunda-
whereas the suspended load is totally entrained into and transported tion process and corresponding erosion and deposition by the 2004 IOT
within the flow. Most forward models separate bedload and suspended at Lhok Nga, in northwest Sumatra, Indonesia, using several snapshots
load transport (Table 1); in some models transport in bedload and of the distributions of flow speed and bed level change. They demon-
suspended load are combined into a total load transport (Table 1; strated that a flow speed on the order of 20 m/s and a flow depth up to
Li et al., 2012a,b). The horizontal and vertical movement of suspended 5 m are responsible for the transport of a large amount of sediment. Ero-
sediment is computed using advection–diffusion equations (e.g., Kantha sion near the coastline occurs during the passing of the wave front, and
and Clayson, 2000). The morphological change due to the erosion and de- deposition occurs after the wave front passed by. A significant amount
position of sediment is calculated using an Exner-type equation (e.g., of sediment is eroded by the backwash, resulting in a net erosion of the
Leliavsky, 1955), which describes the conservation of sediment mass. shoreface (Li et al., 2012b). Other details of the sediment transport pro-
One of the key non-dimensional parameters that describes the cess, such as the spatial and temporal change in suspended sediment,
transport of sediment is the Shield's parameter, which is defined as have been determined by many other studies (Kihara and Matsuyama,
the ratio of the destabilizing fluid force (i.e., bottom shear stress) that 2010; Yoshii et al., 2010; Apotsos et al., 2011b; Gusman et al., 2012;
acts on a sedimentary particle to move it to the stabilizing force due to Kihara et al., 2012; Ranasinghe et al., 2013). For example, the 3D modeling
submerged weight that acts on a particle to resist movement. The bot- of sediment transport by the 2004 IOT offshore of the Kirinda fishery har-
tom shear stress is a tangential force due to the friction between the bor in Sri Lanka revealed that eddies significantly affect the transport and
fluid and the bed, which may initiate motion of sedimentary particles. deposition of suspended sediment (Kihara and Matsuyama, 2010). Their
The investigation of the physical process and dimensional analysis of simulation showed that secondary flow transports suspended sediment
sediment transport have revealed that the amount of sediment toward the center of the eddy, where it accumulated because flow speeds
transported as bedload can be expressed as a function of Shield's param- there are lower than at the outer circumference of the eddy.
eter (e.g., Meyer-Peter and Müller, 1948; Einstein, 1950; Bagnold, 1966;
Ashida and Michiue, 1972; Van Rijn, 1984a). A variety of formulations 3.2.1.2. What processes control the geometry and internal structure of tsu-
have been proposed so far to predict the bedload transport rate. nami deposits? For example, the general landward thinning and fining
Among the formulations that are listed in Table 1, those by van Rijn observed in onshore tsunami deposits is believed to reflect the gradual
(van Rijn, 1984a, 1993; van Rijn et al., 2004) are most widely used. loss of flow energy during tsunami run-up and are used as criteria to as-
Some studies used other formulations in which some empirical param- sign a tsunami origin for deposits (Goff et al., 2012). The sophisticated
eters are derived from flume experiments (Takahashi et al., 2000). The modeling by Apotsos et al. (2011b) showed that the erosion of near-
amount of suspended load can be calculated by specifying vertical diffu- shore sediment by backwash is another important process that controls
sion due to turbulent mixing and suspended sediment concentration at the thickness and grain size distribution of onshore tsunami deposits.
a reference level, namely, the reference concentration. It has been They illustrated sediment transport during the backwash and onrush
known that the reference concentration is a function of the Shield's pa- of the 2004 IOT at Kuala Meurisi, in northwest Sumatra, Indonesia,
rameter (Lane and Kalinske, 1941; Einstein, 1950; Van Rijn, 1984b). The using instantaneous snapshots of flow speed, turbulent energy and sed-
vertical diffusion is determined by the hydrodynamic model. Alterna- iment concentration as well as the spatial structure of the resulting tsu-
tively, some of the models employ formulations that describe the nami deposit. Their simulation showed that sediment seaward of the
explicit relationship between the pick-up rate and the Shield's parame- coastline experienced a long-duration, high-velocity flow due to the
ter to calculate the entrainment of sediment as suspended load (e.g., backwash, and a significant amount of the sediment is suspended in
Takahashi et al., 2000; Yoshii et al., 2010). The settling rate of the the offshore. The collision of the sediment-laden backwash with the on-
suspended load is defined as a product of the settling velocity of the rush by the following wave results in a sudden deposition of sediment
particle and the depth-averaged or reference concentration of the and the formation of large offshore bar. A portion of the offshore
suspended load. A variety of formulations for settling velocity of natural suspended sediment is advected by the onrush and deposited on land
grains that account for grain shape, density, and other factors have been (Apotsos et al., 2011b). They noted that the concentration of suspended
used in previous studies (e.g., Dietrich, 1982; van Rijn, 1984b, 1993; sediment in the onrush may not be in equilibrium with the local hydro-
Soulsby, 1997). dynamics, and onshore deposition may be affected by the large lateral
gradient of the suspended sediment. Note that, under non-equilibrium
3.2. Current applications in tsunami geology conditions, sediment concentration cannot be determined by local
flow depth and speed, and therefore, the resulting deposition may not
Understanding the sedimentary processes of tsunami deposit forma- necessarily reflect the flow condition. Although higher turbulent energy
tion is one of the major interests of tsunami geologists. However, our appears near the front of tsunami inundation, the suspended sediments
understanding of sedimentation processes is largely dependent on in- are mainly concentrated behind it. The simulated tsunami deposit on
terpretation because we have no way to observe actual sedimentation. smooth slopes showed a clear landward thinning and fining trend,
Forward modeling can provide complementary information to better whereas the distribution pattern of the deposit is more complicated
understand the sedimentary processes responsible for tsunami deposi- on the measured topography.
tion and erosion. Here, we summarize studies that use numerical
models to answer fundamental questions of tsunami geologists. 3.2.1.3. How do multiple tsunami waves affect erosion and deposition?. In
general, tsunamis consist of a train of waves with varying amplitudes
3.2.1. Questions stimulated by field observations and periods. During the inundation by each wave, a change or rever-
Field observations of tsunami deposits raise many questions about sal of the flow direction (backwash) takes place. Observations of
how a tsunami moves sediment resulting in erosion and deposition. modern tsunami deposits have documented that they sometimes
Below are a few of the questions geologists have posed that have been consist of multiple sedimentary units. Such a structure can be associ-
addressed using forward numerical models. ated with deposition either by onrush or backwash or even by
following waves (e.g., Dawson and Stewart, 2007; Goff et al., 2012), 3.2.2.2. What are the sediment sources for onshore deposits?. Apotsos et al.
and the behavior of a tsunami has been reconstructed based on the (2011c) investigated the effects of the initial distribution of sediment
interpretation of these units (Nanayama and Shigeno, 2006). The and onshore slope on the deposition of sediment using an idealized
number of episodes of onrush and backwash that are recorded in sediment-limited embayment. In their study, most onshore tsunami
the deposit may depend on the tsunami's hydrodynamic features deposits are transported from the beach and shallow nearshore, regard-
and local topography. For locations inundated by multiple waves, less of where the sediment was initially distributed. Onshore deposition
the following three possible scenarios have been identified using of sediment is reduced by more than 40%–50% by limiting the supply
simulated time series of the water and bed levels (Li et al., 2012b): of sediment and is increased 60%–80% by a moderate onshore slope.
(1) successive waves or backwash totally erode the pre-existing tsu- Li et al. (2012b) carried out a numerical experiment on tsunami sand
nami deposits, (2) each successive wave partly erodes the pre- transport by varying the initial grain-size composition and spatial distri-
existing tsunami deposits, and (3) some of the pre-existing tsunami bution. They showed that the thickness of tsunami deposits is affected
deposits are partly eroded while others are totally removed. Li et al. by both grain size and sediment source; however, the local topography
(2012b) noted that the thickness of the tsunami deposit only exerts greater control on the distribution of tsunami deposits.
records the accumulated effect of flow on the sediment. This means
that the number of units cannot necessarily be linked to the number
3.2.2.3. How are morphological changes induced by a tsunami, and where
of waves.
should we search for paleotsunami deposits?. Li et al. (2012a) investigated
Fig. 2 shows an example of how forward modeling elucidates how
tsunami sediment transport in a small embayment based on six projected
sediment transport during multiple tsunami waves results in net erosion
earthquake-tsunami scenarios. In their modeling, the character of the tsu-
and deposition. In this example, erosion and deposition at 20 and 110 min
nami at the bay mouth, defined as either a positive wave (crest) or a neg-
after the arrival of the tsunami at the shoreline are shown. At 20 min the
ative wave (trough) leading the tsunami waveform, is closely associated
first wave has totally retreated from the pre-tsunami coastline. At
with the mechanism of the earthquake scenario, and the patterns of sed-
110 min the third wave is going to retreat. This example shows
iment transport differ from each other. For a positive leading wave, a large
that the erosion of the beach is mainly explained by the onrush and
amount of sediment is transported seaward by the backwash and forms a
backwash during the first and second waves. The erosion and depo-
large nearshore sand bar. For a negative leading wave, a considerable
sition near the coastline was not significant during the third wave,
amount of sediment is suspended and transported seaward during the
although the flow speed reached up to 2 m/s. Forward modeling en-
initial retreat of the sea. When a positive wave comes into the bay, a sud-
ables us to understand the possible sources and transport paths of
den decrease in flow speed takes place, and some of the suspended sedi-
sediment and provides insight into the processes that form the com-
ment is deposited locally, forming a large offshore sand bar. It is
plex geometry and internal structure of tsunami deposits. Forward
interesting that this formation process of the offshore sand bar is nearly
modeling also allows us to better interpret the behavior of the tsuna-
identical to the results of another study (Apotsos et al., 2011b). They ar-
mi from its deposits and may also be useful for improving identifica-
gued that the difference in the erosion and deposition patterns between
tion and utilization of tsunami deposits.
the two tsunami scenarios provides useful information for tsunami geol-
ogists to further study paleotsunamis in the area.
3.2.2. Knowledge gained from numerical experiments
Complexity and variability of tsunami deposits underscore the
Considering the inherent difficulties in making field measurements
need to better understand the factors controlling the features of
and scaling laboratory experiments of tsunami sediment transport, an-
tsunami deposits (e.g., Goto et al., 2012a). Numerical experiments
other advantage of forward modeling is its capability to conduct numer-
can demonstrate which factors can dominate tsunami sediment
ical experiments to better understand tsunami deposition and erosion.
transport. Given that the underlying physics are satisfactory incor-
Simulations with varying external forcing, topographic settings and
porated into the modeling, the modeled relationship between the
sediment sources may aid to in the understanding of what factors con-
key factors and the deposit may clarify which data is important
trol tsunami sediment transport, why tsunami deposits are so diverse,
for reconstruction of the tsunami event, as shown by Li et al.
and what characteristics of tsunami deposits can be used to discriminate
(2012a). However, the relationships learned from the modeling
them from other extreme wave event deposits. Below are several of the
cannot be exploited if the essential field data are not collected.
many questions tsunami geologists ask that have been addressed using
We may not be successful in accurately assessing the tsunami
numerical experiments.
hazard. Therefore, there is an impetus to conduct sedimentary re-
search to better constrain the key factors that control how tsunami
3.2.2.1. How can we differentiate tsunami and storm deposits?. Although
hydrodynamics relate to deposit features. Numerical experiment
numerous investigations and discussions have been published on differ-
may provide valuable insights into what field data needs to be col-
entiating tsunami and storm deposits (e.g., Kortekaas and Dawson,
lected and the interpretation of tsunami hydrodynamics (both
2007; Morton et al., 2007), universal diagnostic criteria have not yet
paleo and modern) from tsunami deposits.
been established. The most significant differences in the hydrodynamics
of tsunamis and storms derive from differences in their wave ampli-
tudes and wavelengths. Forward modeling may provide insight into 3.3. Current issues and future challenges
this issue. Gusman et al. (2012) showed a series of simple 1DV numer-
ical experiments of tsunami sediment transport on varying ideal topog- 3.3.1. Model improvement
raphy and wavelength. Their simulation illustrated that a tsunami with The accuracy and resolution of forward modeling depends on how
a shorter wavelength caused more erosion near the coastline than one well the model reproduces the physical processes of tsunami hydrody-
with a longer wavelength, given an equivalent wave amplitude for namics and sediment transport. Authors of previous publications have
each tsunami scenario. This can be explained by the fact that the tsuna- emphasized the importance of including the details of the physical pro-
mi energy of the shorter wavelength induced larger flow acceleration cesses for simulations to resemble reality. Below is a list of challenges to
than that of the longer wavelength (Gusman et al., 2012). The pattern further improvement of numerical models of tsunami sediment trans-
of erosion and deposition are accentuated by the change of onshore port that were addressed by the previous studies. Note that some
slope. In their numerical experiment, the waveform represented only numerical models already have implemented some of these improve-
tsunamis. An experiment using waveforms that represent storms is a ments, such as (5) three-dimensional computation (Kihara and
natural extension, and it may prove fruitful for sedimentological re- Matsuyama, 2010), (6) mixed grain size (Apotsos et al., 2011b, Li et al.,
search on tsunamis and storms. 2012a,b) and (8) infiltration and exfiltration (Nakamura et al., 2009).
1. Applicability of available formulations of sediment pick-up rate 2012); however, sedimentary data has not yet been obtained together
2. Bottom friction and Manning's roughness with estimated flow speed.
3. Change in tsunami hydrodynamics due to increasing debris during Note that although tsunamis may cause an exchange of sediment
inundation and other types of materials across the coastline, a comprehensive
4. Effect of turbulence, such as due to wave breaking dataset that includes data from both the sea and the land does not cur-
5. Three-dimensional sediment transport formulations for complex to- rently exist. At present, only a few datasets from either the sea or the
pographic settings land are available to benchmark models (Table 2; see Appendix D). Con-
6. Mixed grain size sidering the complexity of the hydrodynamics of a tsunami and the var-
7. Lateral variation of flow iability of tsunami deposits, the existing benchmarks are insufficient to
8. Pore-water pressure gradient and infiltration and exfiltration of adequately validate models. In this regard, the massive dataset from the
water from the surface 2011 Tohoku-oki earthquake tsunami, introduced in Section 6, may ad-
vance validation of numerical models for tsunami sediment transport.
Of these, the first two suggested improvements may play important
roles in accurate modeling of tsunami sediment transport (see
Appendix A). As noted by Apotsos et al. (2011a), the complexity and un- 4. Inverse modeling of sandy tsunami deposits
certainty of coupled tsunami modeling are increased due to the inclu-
sion of sediment transport. Further implementation of the physical Tsunami deposits capture the details about the waves that formed
processes listed above may improve the model's capabilities but also them — we just have to learn how to read this information. Inverse
further increase the complexity and associated uncertainty through modeling is a quantitative approach for interpreting tsunami character-
the inclusion of additional empirical or even undetermined model pa- istics, such as tsunami flow speed and height, from deposits. The
rameters. With increasing complexity, there is a greater need to opti- starting point for development of inverse models is identifying key fea-
mize the set of empirical parameters for the models to perform well. tures of the tsunami deposits that may include deposit thickness, hori-
zontal and vertical variation in grain size, contacts, and structure. The
3.3.2. Model validation question asked is what tsunami characteristics created the observed
A comprehensive validation study of existing numerical models is features. The inverse approach holds great potential for improving tsu-
needed before we can trust their results when applied to paleotsunami nami hazard assessment when applied to paleotsunami deposits. For
deposits (Cheng and Weiss, 2013). The importance of benchmarking example, tsunami hazard on the Sendai coastal plain, which was
numerical models of tsunami sediment transport and the desirable at- underestimated because of an incomplete understanding of the magni-
tributes of a benchmark are discussed by Huntington et al. (2007). The tude (height, speed, and inundation) of past tsunamis, likely would
comprehensive dataset of a benchmark may include sedimentary data, have been improved by application of current state of the art of inverse
such as thickness, grain size, and structure, as well as information on models to known paleotsunami deposits.
the tsunami hydrodynamics, such as the number of waves, flow depth
and speed, and flow indicator. Pre- and post-tsunami topography and
vegetation cover may also be part of a benchmark. Issues regarding 4.1. Current state of the science of inverse modeling of sandy tsunami
the validation of numerical modeling are tightly linked with the deposits
model setup and resolution. Because typically validation is performed
using recent tsunami events as benchmarks, the discussions here are The published inverse models for interpreting sandy tsunami de-
mainly focused on the availability of data. The resolution and precision posits start with a basic assumption that tsunami deposits are formed
of the topography data may have a considerable effect on model results from sediment settling out of suspension (Table 3) (Jaffe and
and on comparison to benchmark data (see Appendix B). Data on sur- Gelfenbaum, 2007; Moore et al., 2007; Smith et al., 2007; Soulsby
face features (e.g., distribution of buildings, open areas, etc.), such as et al., 2007). This simplification will not hold when the tsunami size is
can be derived from a land use map, is recommended in a benchmark small and corresponding flow speeds are too low to suspend sediment
to evaluate the ability of a model to reproduce surface-feature effects or for all parts of a tsunami deposit (Moore et al., 2011; Jaffe et al.,
on tsunami hydrodynamics and sediment transport (see Appendix C). 2012). However, a prevalent observation is that tsunami deposits fine
In general, numerical models should not only be validated on their upward (normal grading) (Shi et al., 1995; Minoura et al., 1997;
ability to reproduce features of the observed deposits, but also on how Gelfenbaum and Jaffe, 2003; Jaffe et al., 2003; Jaffe et al., 2006;
well they reproduce the temporal evolution of deposit formation. The Razzhigaeva et al., 2006; Szczuciński et al., 2006; Moore et al., 2006;
lack of in situ measurement of tsunami sediment transport makes it dif- Hawkes et al., 2007; Hori et al., 2007; Morton et al., 2007; Choowong
ficult to validate numerical models (Apotsos et al., 2011a). Given their et al., 2008; Bourgeois, 2009 and references therein; Fujino et al.,
relevance to bottom shear stress, measured or estimated flow speeds 2010; Peters and Jaffe, 2010 and references therein). Fining upward is
are indispensable for model validation, and many authors have ad- consistent with deposition from suspension, and some studies (e.g.,
dressed the importance of agreement of flow speed between the simu- Jaffe et al., 2012; Witter et al., 2012) have shown that the observed up-
lation and reality. Accurate flow depths and speeds are essential for ward fining can be reproduced by modeling of sediment settling out of
forward modeling of sediment transport (Ontowirjo et al., 2012). Agree- suspension from the water column. So, it is reasonable to develop in-
ment in tsunami heights and flow depths does not necessarily mean a verse models for sandy tsunami deposits based upon deposition from
model is able to accurately reproduce the temporal variation of the suspension. In addition, all of the inverse models are for onland tsunami
flow speed (Apotsos et al., 2011a). Recently, and particularly since the deposits.
2004 IOT, flow speeds during tsunami inundation have been estimated In the remainder of this section, each of the inverse modeling ap-
by analyzing video records (Fritz et al., 2006; Hayashi and Koshimura, proaches is presented. Emphasis is placed on the assumptions of the
Fig. 2. Example of the time evolution of tsunami-induced deposition (a and b), erosion (c and d) and (e) time series of ground elevation, water level and flow speed on the pre-tsunami
coastline (at point P1) simulated by a forward numerical model. Temporal changes in the maximum deposition (a and b) and maximum erosion (c and d) between two different timings
are illustrated. The blue arrow indicates the location of significant deposition 110 min after the tsunami attack. The red arrow indicates the location where significant erosion takes place
110 min after the tsunami attack. The yellow star shows the location of the time series. Note that the area illustrated in this figure is a part of Figs. 9–12. Details of the model are the same as
in Fig. 12.
Table 2
Available datasets used for the forward numerical modeling of tsunami sediment transport.
Location Event Morphological data Waveform Inundation Tsunami Flow References

from area depths speed
nearby site and heights
Pre- Post-
tsunami tsunami
Sea Land Sea Land
Kesennuma 1960 ✓ ✓ ✓ ✓ Takahashi et al. (2000)

Chile
Papua New 1998 PNG ✓ ✓ ✓ ✓
Guinea Titov et al. (2001), Gelfenbaum and Jaffe (2003), Huntington et al.
(2007)
Kirinda 2004 IOT ✓ ✓ ✓
Nishihata et al., (2006), Takahashi et al. (2008) Kihara et al.and
Matsuyama (2010)
Lhok Nga 2004 IOT ✓ ✓ ✓
Li et al. (2012a,b), Ontowirjo et al. (2012), Gusman et al. (2012)
Kuala Meurisi 2004 IOT ✓ ✓ ✓ Apotsos et al. (2011a,b)
Fagafue Bay 2009 ✓ ✓ ✓ Apotsos et al. (2011c)
Samoa
models to allow evaluation of what environments and conditions they not resuspended after deposition (i.e., a single trajectory), and (3) the
may be applicable. large particles are suspended throughout the water column and present
near the water surface (tsunami height). With these assumption, the
4.1.1. Particle trajectory model time, t, for a particle near the water surface at the shoreline (assumed
The particle trajectory model by Moore et al. (2007) is based on the source location) to settle to bed where it is deposited (LHS of Eq. (1))
observation that tsunami deposits tend to fine inland. The concept for is equal to the time to travel inland (RHS of Eq. (1)) to where it is depos-
the model is that the trajectory (and distance) that a particle travels ited as given by
moving inland to the location where it is deposited is determined by
its settling velocity and the tsunami flow speed (Fig. 3a). Coarser parti- H l
¼t¼ ð1Þ
cles do not travel as far inland because they settle out of the water more ws U
quickly than finer particles. The coarser the particle, the closer it will be
deposited to the place where it is picked up. There are several key as- where H is tsunami height at the shoreline, ws is settling velocity, l is the
sumptions (Table 3) in the particle trajectory model of Moore et al. distance traveled by the particle, and U is the average speed of the par-
(2007): (1) the particle is transported in suspension, (2) the particle is ticle from the shoreline to the where it is deposited.
Table 3
Inverse approaches for determining tsunami characteristics from deposits.
Approach Tsunami characteristics estimated Inputs Assumptions References
Particle trajectory Product of tsunami height at shore and Settling velocity of a larger − Transport in suspension Moore et al. (2007)
average speed (discharge), can use Law of (D90 or D95) particle, − Particle not resuspended after
the Wall or Froude number assumptions distance particle traveled deposition (single trajectory)
to calculate each parameter separately − Large particles suspended
throughout the water
− Law of the Wall applies
(2007 paper)
Particle settling Minimum depth of water Settling velocity of slowest − Transport in suspension Smith et al. (2007)
settling particle − Particles settle individually
− Muds settle as flocs
− Wave period of tsunami
can be estimated
Settling column Inundation and runup Bulk distribution of settling velocities − Transport in suspension Soulsby et al. (2007)
at multiple locations, fractional thickness − Wave heavily, uniformly,
of grain size components charged with sediment
at the shoreline
− Particles not resuspended after
deposition (single trajectory)
Equilibrium suspension Shear velocity at a point, tsunami flow speed Bulk distribution of settling velocities of − Transport in suspension Jaffe and Gelfenbaum (2007),
at a point calculated using bottom roughness suspension graded interval, bottom − Steady flow formulation for Jaffe et al. (2012)
(high sensitivity to choice) and flow depth roughness, flow depth turbulence
(low sensitivity to choice) estimates (not a strong constraint) − Sediment available for
suspension
− Equilibrium suspension
− Portions of deposit
formed from clearing
of the water column
− No erosion by backwash
flow speed, 1.9–2.2 m/s, were minimums and compared well with eye-
witness observations.
4.1.2. Particle settling model

Smith et al. (2007) use the settling rate for fine sand particles and the
wave period to estimate the minimum depth of water during the
~ 8000 ybp Storegga Slide tsunami. The assumptions (Table 3) for the
model are: (1) transport in suspension, (2) particles settle individually,
(3) muds settle as flocs, and (4) wave period of the tsunami can be es-
timated. Using an estimated wave period of 60 min based on modeling
by others (Harbitz, 1992; Bondevik et al., 2005), they estimated that the
water depth at Montrose, Scotland during the ~8000 ybp Storegga Slide
tsunami was 7.2 m. This is a maximum because settling may not have
occurred for the entire 60 min of the wave period.
4.1.3. Settling column model

Soulsby et al. (2007) also developed an inverse model based on the
observation that tsunami deposits tend to fine inland. Like the Moore
et al. (2007) model, it is based on particle trajectory. Unlike Moore
et al. (2007), which is for uprush only, the Soulsby et al. (2007) model
includes deposition by both tsunami uprush and backwash. The concept
for the model is that sediment is treated as if it were a moving settling col-
umn (Fig. 4a). The tsunami deposit thins and fines landward as the result
of settling of sediment, with coarser particles settling faster (closer to the
shoreline) than finer ones. Deposition can also occur on backwash. Unlike
Moore et al. (2007) where a single particle is tracked, the entire range of
particle sizes are tracked. Tracking of water and particle trajectories result
in expressions for the landward variations in thickness and sediment
grain size of the deposit. The key assumptions (Table 3) of the Soulsby
et al. (2007) approach are: (1) particles are transported in suspension,
(2) the wave is heavily charged with a vertically-uniform concentration
of sediment at the still water line [shoreline], and (3) the particles are
not resuspended after deposition [i.e., single trajectories].
The inland extent of sediment, Is, is related to the inland extent of
Fig. 3. Particle trajectory model of Moore et al., 2007, showing a) model concept of the trajec-
water, Iw, by
tory (and distance) that a particle travels moving inland to the location where it is deposited
is determined by its settling velocity and the tsunami flow speed, and b) application of the
Iw
model to the 1929 Grand Banks tsunami in Newfoundland. Shaded regions are unlikely solu- Is ¼ ð3Þ
tions. The tsunami height and average flow speed predicted by the model is at the intersec- 1 þ αγ
tion of the slanting line and the Law of the Wall curve, which give more information and
allows separating the product of tsunami height and average flow speed into individual com-
ponents (modified after Moore et al., 2007).
where γ is the fraction of the wave period during which uprush occurs,
and
Rearranging Eq. (1) gives ws T

α¼ ð4Þ
H
HU ¼ lW s : ð2Þ
where T is period.
The thickness of the tsunami deposit at the still water line (shore-
The settling velocity and distance traveled by the particle are mea- line), ζ0, is
surable quantities. Their product (RHS Eq. (2)) is equal to two unknown
parameters, the product of the tsunami height and average speed of the α ð1 þ αγ Þ C 0
ζ0 ¼ H ð5Þ
particle (LHS Eq. (2)) along its journey from the shoreline to the location 1þα ρB
where it is deposited. The LHS of Eq. (2) is also a discharge. Reinhardt
(1991, unpublished Masters Project available from the University of where C0 is the concentration of suspended sediment in the water col-
Washington Geology Library) presented similar concepts and other ap- umn (vertically uniform), and ρB is the dry bulk density of the deposited
proaches for interpreting hydrodynamic conditions during the 1700 sediment.
Cascadia tsunami in coastal Washington from its deposits. The spatially varying thickness of the tsunami deposit is
To solve Eq. (2) (two unknowns) requires another equation. Moore
et al. (2007) used the relationship between flow speed and depth as-
x
suming that the flow could be approximated by a unidirectional flow ζ ðxÞ ¼ ζ − 1− for x≤ Is ; and 0 for xNI s ð6Þ
Is
with a fully developed boundary layer (Law of the Wall). With this ad-
ditional information it is possible to estimate the flow speed and
depth independently based on the measured settling velocity and dis- where x is distance inland from the still water level (shoreline).
tance the particle traveled inland. Soulsby et al. (2007) apply Eqs. (4)–(6) separately to different grain
Moore et al. (2007) applied the particle trajectory model to a deposit size fractions. The result is that each grain size contribution to the de-
in Newfoundland, Canada from the 1929 Grand Banks tsunami (Fig. 3b). posit thins linearly landwards, with the inland limit of sediment increas-
They concluded that the resulting average flow depth, 2.5–2.8 m, and ing with decreasing grain size. Analysis of different grain size fractions
Fig. 4. Settling column model of Soulsby et al., 2007, showing a) concept of treating sediment as if it were a moving settling column. The tsunami deposit thins and fines landward as the
result of settling of sediment, with coarser particles settling faster (closer to the shoreline) than finer ones. The schematic connotes deposition during uprush; although, the model includes
deposition during backwash as well, and, b) application of the model to the 1929 Grand Banks tsunami in Newfoundland. The inland limits of sediment deposition for grain size fractions
are fitted. The fits are used to calculate the limit of inundation the runup height (modified after Soulsby et al., 2007).
allows a comparison to field data that gives multiple results that can be The model determines the flow speed necessary to suspend the
compared to each other to test for consistency. sediment deposited in a normally graded interval. Jaffe et al. (2011,
Soulsby et al. (2007) give two examples of the application of the the- 2012) observed that a distinctive type of normal grading, termed
ory to field data — the 1929 Grand Banks tsunami in Newfoundland, suspension-grading, is present in intervals of tsunami deposits where
Canada and the ~ 8000 ybp Storegga Slide tsunami in Montrose, the entire distribution shifts to finer sizes moving upward in a deposit.
Scotland. Eqs. (4)–(6) were fitted to observed sedimentary characteris- Intervals of turbidites, another deposit formed by high flow speeds,
tics to deduce the run-up height and the inundation distance of the are also suspension graded (Kuenen and Menard, 1952; Middleton,
water. The estimated values for runup in Newfoundland is 7.4 m 1967, termed this type of grading “distribution grading”). Suspension
(Fig. 4b), which is less than, but close to the observed value 8.8 m. The grading is found at the top of layers of a tsunami deposit and interpreted
estimated runup at Montrose is 6.3 m, which is similar to the 7.2 m to form at the end of each wave uprush when tsunami flow speed de-
value for water depth that Smith et al. (2007) estimated based on creases and suspension can no longer be maintained by lower turbu-
what is most-likely the same data set. Runup and water depth are not lence levels causing clearing of the water column and deposition. The
exactly equivalent, but may be similar in this setting. model calculates multi-class sediment suspension and assumes local
equilibrium between turbulent suspension and settling. This model is
4.1.4. Equilibrium suspension model applied at a point and only to intervals that are deposited from suspen-
Jaffe and Gelfenbaum (2007) developed an inverse model based on sion (Jaffe et al., 2012).
the observation that tsunami deposits commonly fine upward (normal Standard sediment transport formulae are used to determine the
grading). The concept for the model is that sediment in suspension, equilibrium concentration profile for each size class. A Rouse-type ex-
which is in equilibrium with the tsunami flow speed, is deposited pression (formulation in Jaffe and Gelfenbaum, 2007) is used to calcu-
when the tsunami slows (Fig. 5a). A normally graded interval in the tsu- late sediment concentration in the water column that results from
nami deposit is formed during the uprush of each tsunami wave. The upward diffusion in balance with downward settling of grains. This re-
sediment in each normally graded interval is equivalent to the sediment sults in an exponential decrease in suspended sediment concentration
that was in equilibrium suspension in the water column before the tsu- with distance from the bed with a greater decrease for higher settling
nami slowed. The key assumptions (Table 3) of the Jaffe and velocities (sediment tends to be lower because it settles faster) and
Gelfenbaum (2007) approach are: (1) transport in suspension, lower flow speeds that generate less turbulence for the same bottom
(2) steady flow formulation for turbulence is applicable to the phase roughness (sediment tends to be lower because the energy to mix it
of a tsunami near the maximum flow speed, (3) sediment is available up into the water column is less).
for suspension [i.e., a deposit will form], (4) suspended sediment is dis- The model iteratively adjusts the input sediment source distribution
tributed in the water column in equilibrium concentration profiles, and shear velocity (U*, a parameterization of sediment pick-up and tur-
(5) portions of the deposit formed from clearing of the suspended sed- bulent mixing intensity) to match the size distribution and amount of
iment in the water column, and (6) no erosion by backwash. sediment in suspension to that of a suspension-graded interval of a
layer of the tsunami deposit (see Jaffe et al., 2012 for details). This is problem, because bottom roughness is even more unknown than for
done for each size class, i, as modern tsunamis, the translation from shear velocity to tsunami flow
speed is a larger uncertainty. Witter et al. (2012) likely underestimated
z
Zh Z
SGLtop bottom roughness like Jaffe and Gelfenbaum (2007) did, with the same
C i ðzÞdz ¼ C i ðzÞdz ð2Þ result of a likely overestimate of tsunami flow speed. Had a Manning's n
0 z
SGLbot of 0.03 parameterization been used, flow speeds would have been com-
parable to those calculated by a 2-D hydrodynamic inundation model
where Ci(z) is the sediment volume concentration of the size class i at (Witter et al., 2012).
elevation z above the bed, zSGLbot is the bottom of a suspension-
graded interval of the tsunami deposit and zSGLtop is the top of a 4.2. Research challenges for inverse modeling of sandy tsunami deposits
suspension-graded interval of the tsunami deposit.
After determining the shear velocity needed to produce a suspension- Inverse modeling of tsunami deposits to learn tsunami characteris-
graded interval of a deposit, the flow speed profile, U(z) is calculated by tics is a relatively new area of research, with most of the work being
done in the past decade. As such, all of the models have yet to be thor-
Zz oughly tested (and as a result, most-likely improved). Of the four
U 2
U ðzÞ ¼ dz ð3Þ models presented in this section, only one, Jaffe and Gelfenbaum
K ðzÞ
z0 (2007), has been tested using deposit and tsunami characteristic
(height, speed, etc.) data from recent tsunamis. Two others, Moore
where z0 is the bottom roughness and K(z) is the eddy viscosity. This et al. (2007) and Soulsby et al. (2007), have been tested using data
method for calculating the flow speed profile is standard (e.g., Van Rijn, from historic tsunamis. There is data from recent tsunamis that can be
1993). used to test all of the models presented. This should be done.
The vertical variation in grain size distributions of a suspension- Two of the models presented, Smith et al. (2007) and Jaffe and
graded portion of the tsunami deposit is reconstructed and compared Gelfenbaum (2007), have been applied to paleotsunamis. The impor-
to the actual distributions of the tsunami deposit sampled in the field. tance of testing the models on paleotsunami deposits is twofold: (1) it
The reconstruction follows the methods of Jaffe and Gelfenbaum is paleotsunamis characteristics [magnitude] that ultimately we are try-
(2007). The reconstruction is simple — sediment in suspension is ing to better understand to improve tsunami hazard assessment, and
allowed to settle and the amount settling in each size class is tracked (2) not all of the models may work on deposit that have been altered
as the deposit accretes (see appendix in Jaffe et al., 2012 for details). If by bioturbation, weathering, and other post-depositional processes.
the grading of the reconstructed deposit matches that of the observed Testing inverse models on paleotsunamis is more challenging because
deposit reasonably well then the assumption that the sediment was de- model inputs may be more difficult to specify and independent mea-
posited from suspension is supported and the application of the model sures of tsunami characteristics may be more equivocal. Yet, because
is likely appropriate (Jaffe and Gelfenbaum, 2007). the ultimate goal of this research is to improve tsunami hazard assess-
The Jaffe and Gelfenbaum (2007) inverse model has been applied to ment, tests and methods should be developed to discern the accuracy,
deposits from four recent tsunamis (Jaffe and Gelfenbaum, 2007; Spiske limitations, and applicability of inverse modeling of tsunami deposits.
et al., 2010; Jaffe et al., 2011, 2012) and one paleotsunami (Witter et al., A grand challenge for modeling of tsunami deposits is to develop
2012). For recent tsunamis, tsunami flow speeds either measured or es- methods for combining inverse and forward modeling to increase the
timated using other approaches (e.g., forward hydrodynamic model) information that can be extracted from the deposit. Development of hy-
are similar to those estimated using the Jaffe and Gelfenbaum (2007) brid approaches for extracting tsunami flow speed and other informa-
model. For example, for the 2011 Tohoku-oki tsunami on the Sendai tion from deposits that combines forward and inverse sediment
coastal plain near the Sendai airport, modeled maximum flow speeds transport and hydrodynamic models holds great promise for unlocking
at 1149 m and 1344 m inland from the shoreline are similar (4.2– more information from paleotsunami deposits and improving tsunami
4.3 m/s) to the average speed of the tsunami wave front (4 m/s, Goto hazard assessment.
et al., 2011a) calculated from video collected from a helicopter of the
first wave as it traveled from 1.1 to 2.1 km from the shoreline (Fig. 5b; 5. Boulder transport models
see Jaffe et al., 2012 for additional comparisons). This comparison is
based on using a bottom roughness parameterized by a Manning's n 5.1. Recent progress in tsunami boulder transport research
of 0.03, which is appropriate for the sand and grass substrate present
(Sugawara and Goto, 2012), to calculate tsunami flow speed from Boulder deposit research is a growing field in tsunami geology
shear velocity. The importance of the choice in bottom roughness can- (e.g. Scheffers and Kelletat, 2003; Goto et al., 2010a; Switzer and
not be overemphasized. An interesting illustration of this is that the Burston, 2010; Etienne et al., 2011; Paris et al., 2011). Here we use
use of an unrealistically low bottom roughness, one only based on the term boulder to describe clasts larger than 25.6 cm. Similar to
grain roughness and a saltating bedload layer, in Jaffe and Gelfenbaum sandy tsunami deposits, boulder deposits are expected to provide
(2007) resulted in likely overestimate (~ 14 m/s) of the tsunami flow useful information to better understand the histories, recurrence in-
speed for the 1998 Papua New Guinea tsunami at Arop. This flow tervals, and size of paleotsunamis. Significant efforts have been made
speed was similar to one calculated by a 1-D hydrodynamic model to establish the identification criteria for boulders of tsunami origin
that was frictionless (certainly too low a bottom roughness) (Titov to discriminate them from other extreme event (e.g. storm waves)
et al., 2001). Had a more realistic bottom roughness, for instance a boulder deposits and to reconstruct the flow speed and height of
Manning's n of 0.03, been used, the flow speed calculated by the Jaffe the tsunami that transported the boulders (e.g., Nott, 2003; Goto
and Gelfenbaum (2007) model would have been 50% or more lower. et al., 2010c; Nandasena et al., 2011b).
Spiske et al. (2010) state in their abstract, “The testing of TsuSedMod Until the early 2000s, many studies focused on exploring the origin
on recent examples shows it to be a valuable tool for calculating flow of enigmatic coastal boulders regardless of whether they were deposit-
depths and speeds of ancient tsunami.” Witter et al. (2012) applies ed by a tsunami or storm wave (e.g., Young and Bryant, 1992; Young
the inverse model to paleotsunami deposits from the 1700 Cascadia tsu- et al., 1996; Nott, 1997). The 2004 IOT was an important turning point
nami at Cannon Beach, Oregon. Several weaknesses in the inverse because boulder deposits were found in places where the tsunami hy-
model became apparent in this study. The greatest weakness of the drodynamic features and local topography is known, and pre- and
model is the uncertainty in bottom roughness. For the paleotsunami post-tsunami satellite images are available. Thus, their sedimentary
processes have been well studied using field observations and numeri- establish their tsunami as opposed to storm origin, and to estimate the
cal modeling (Goto et al., 2007, 2010b; Kelletat et al., 2007; Paris et al., hydrodynamic features of the event from the deposited clasts.
2009, 2010; Yawsangratt et al., 2009). Boulder deposits formed by other extreme events such as storms
After the 2004 IOT, a great deal of attention was paid to boulder de- were also studied extensively because understanding their sedimentary
posits during post-tsunami field surveys, and there has been a signifi- process is very important for differentiating boulders deposited by tsu-
cant increase in the number of papers about boulder deposits, namis from other extreme events such as storms (e.g., Onda, 1999;
including many artificial objects, formed by the recent tsunami events. Mastronuzzi and Sansò, 2004; Noormets et al., 2002; Nott, 2004;
These studies include ones on the 1992 Nicaragua tsunami (Higman Scheffers and Scheffers, 2006; Goto et al., 2009a, 2011b, 2013;. Suanez
and Bourgeois, 2008), 2006 central Kuril Islands tsunami (Bourgeois et al., 2009; Etienne and Paris, 2010; Stephenson and Naylor, 2011).
and MacInnes, 2010), 2009 South Pacific tsunami (Lamarche et al., Boulders deposited by historical tsunamis have been found at the
2010; Etienne et al., 2011; Richmond et al., 2011a), 2010 Chilean tsuna- Ryukyu Islands in Japan (Kato and Kimura, 1983; Kawana and Nakata,
mi (Morton et al., 2011; Spiske and Bahlburg, 2011) and the 2011 1994; Goto et al., 2010a,c) and Hawaii in the USA (Goff et al., 2006;
Tohoku-oki tsunami (Goto et al., 2012b; Nandasena et al., 2013) Richmond et al., 2011b). Estimating the tsunami source model and the
(Fig. 6a). As Goto et al. (2012b) mentioned, the major goals of these recurrence interval using boulder deposits is challenging (Goto et al.,
studies were to describe the sedimentary features of the boulders, to 2010a). For example, the coral boulders deposited by the 1771 Meiwa
Fig. 5. Equilibrium suspension model of Jaffe and Gelfenbaum, 2007, showing a) concept a suspension-graded interval at the top of a layer formed during the uprush of a tsunami wave. The
sediment in the suspension-graded interval is equivalent to the sediment that was in equilibrium suspension in the water column before the tsunami slowed, and b) application to tsunami
deposits formed during the 2011 Tohoku-oki tsunami near the Sendai airport (Jaffe et al., 2012). The model calculates the shear velocity necessary to pick-up and suspend the grain sizes
observed deposited in the suspension-graded interval. Flow speed is calculated from the shear velocity using a specified bottom roughness (modified after Jaffe et al., 2012).
Fig. 6. (a) A huge concrete block deposited by the 2011 Tohoku-oki tsunami at Miyako City, Japan. Note the scale of the human in front of the block. A flow depth of over 20 m can be
estimated based on the water mark at the cliff. (b) Boulders deposited at Miyara Bay of Ishigaki Island. Some boulders in the bay were deposited by the AD1771 Meiwa tsunami according
to radiocarbon dating (e.g., Kato and Kimura, 1983).
tsunami (Fig. 6b) that devastated the southern Ryukyu Islands are re- understand the difference between cliff-top boulders deposited by a
markable because the deposition of these boulders was recorded in his- tsunami or storm wave.
torical documents (e.g., Goto et al., 2010a). The spatial distribution of
boulders can be used to define the area devastated by the tsunami 5.2. Numerical models for boulder transport
(Goto et al., 2010a). Radiocarbon dating was applied to the ~100 Porites
coral boulders interpreted to be deposited by tsunamis (Suzuki et al., 5.2.1. Background
2008; Araoka et al., 2010, 2014), resulting in an estimated recurrence in- As stated above, field evidence of boulders deposited by known tsu-
terval of approximately 150–400 years for large tsunamis in the south- namis is increasing, and such research has significantly contributed to
ern Ryukyu Islands. our understanding of the sedimentary processes associated with boul-
Cliff-top boulders that were deposited on cliffs that are several tens der deposition and to the interpretation of clasts laid down by
of meters high are also reported on some coasts (e.g., Hall et al., 2006, paleotsunamis (Goto et al., 2010b, 2012b; Etienne et al., 2011; Paris
2010). Although there are some reports that describe cliff-top boulders et al., 2011; Goff et al., 2012). However, the determination of the origin
deposited by known storm wave events (Kato et al., 1991; Fichaut and (tsunami vs. storm waves) of coastal boulders around the world is still
Suanez, 2011; Goto et al., 2011b; Hall, 2011), there are no known tsuna- controversial (e.g., Williams and Hall, 2004; Scheffers et al., 2005;
mi boulders deposited on the top of cliffs. This makes it difficult to Kelletat, 2008; Morton et al., 2008; Engel and May, 2012). Moreover,
even for boulders deposited by known tsunamis, it is difficult to quanti- 2008) that can be obtained from field observations and laboratory
tatively estimate tsunami hydrodynamics using only field evidence. To measurements.
overcome these difficulties, developing numerical models for boulder Minimum flow velocity for initiation of motion can be estimated once
transport has been required. the boulder size, weight, and transport mode are known based on simple
Because the size of boulders (usually larger than 1 m in published re- hydrodynamic equations. These equations require specifying whether
ports) is relatively large compared to the flow boundary layer, the equa- transport is by sliding or overturning and depend on whether the boulder
tions that can be applied to the boulders are different from those for is submerged or subaerial, and whether it is free to move or joint bound-
sandy deposits. Basically, the hydraulic force of the wave acting on the ed (e.g., Nott, 1997, 2003; Normets et al., 2004). To estimate the mini-
boulder can be expressed as the sum of the drag, inertial, lift, friction mum flow depth necessary to move the boulder from the calculated
and reduced gravity forces (e.g., Noji et al., 1993; Nott, 2003; Imamura minimum flow velocity, an assumed Froude number (Fr) is used (e.g.,
et al., 2008; Nandasena et al., 2011a,b). Similar to sandy deposits, Nott, 2003). However, Fr is largely variable, ranging from around ~0.6 to
there are forward and inverse approaches for the numerical modeling 2.0 for tsunamis, depending on the inland distance and tsunami size
of boulder transport. (e.g., Matsutomi and Okamoto, 2010); for storm waves the value is un-
certain. Clearly, determining the appropriate Fr is not straightforward.
5.2.2. Forward model
According to Imamura et al. (2008) and Nandasena and Tanaka 5.3. Challenges for future research
(2013), a forward model is defined as hydrodynamic equations includ-
ing non-linear shallow water (or non-linear long wave) equations. Sim- 5.3.1. Numerical approach to discriminate between tsunami and storm
ply speaking, the hydrodynamic force of the tsunami and the properties wave boulders
of boulders are the given parameters, and the forward model solves Discrimination of boulders deposited by tsunamis from other ex-
whether boulders are moved by the flow (Noji et al., 1993; Imamura treme waves such as storm waves is one of the primary issues of the
et al., 2008; Nandasena et al., 2011a; Nandasena and Tanaka, 2013). field (e.g., Nott, 1997, 2003). The existing inverse models can estimate
Currently, there are two forward models for the transport of boul- the minimum flow velocity necessary to move a boulder (Nott, 2003;
ders by tsunamis, one proposed by Imamura et al. (2008) and the Noormets et al., 2004; Nandasena et al., 2011b). However, inverse
other by Nandasena et al. (2011a). A major shortcoming of these for- model results are equivocal for discriminating event type because the
ward models is that the transport mode, which can be sliding, rolling, estimated minimum flow velocity necessary to transport a boulder
and saltation, is assumed (Imamura et al., 2008). To overcome this diffi- can sometimes be explained by either a tsunami or storm wave
culty, Imamura et al. (2008) introduced an empirical variable coefficient (Barbano et al., 2010; Buckley et al., 2012). In fact, estimates from in-
of friction by assuming that the coefficient decreases concomitantly verse models may significantly underestimate the actual flow depth
with decreased ground contact time when the boulder is transported and velocity in some cases because larger boulders, which could be
by rolling or saltation. Nandasena et al. (2011a) developed a model transported, are not present in the source region. The original spatial
based on a theoretical solution for the transport mode. Both model pa- and grain size distributions of boulders at the source strongly influenced
rameters were tuned based on well-controlled water tank experiments the final distributions (Goto et al., 2010c).
(Imamura et al., 2008; Nandasena and Tanaka, 2013), and the models Depending on the application, forward models have now come into
were applied to field-scale phenomena (Imamura et al., 2008; Goto “practical” use. Although forward models require many further im-
et al., 2009b, 2010b; Nandasena et al., 2011a, 2013). provements (Imamura et al., 2008; Nandasena et al., 2011a), they are
Forward models are used to understand the sedimentary processes useful to better understand the transport and sedimentary processes
of boulder transport for recent tsunamis such as the 2004 IOT (Goto of boulders in terms of tsunami flow behavior (e.g., Goto et al.,
et al., 2009b, 2010b; Nandasena et al., 2011a) and the 2011 Tohoku- 2010b). Moreover, analysis along a shore-normal transect is useful to
oki tsunami (Nandasena et al., 2013). Because accurate flow informa- understand the general relationships between the sizes of boulders
tion is required to use forward models for boulder transport, it is impor- and their transport distances and modes (Goto et al., 2009b;
tant to know the incident wave parameters as the initial condition. Nandasena et al., 2011a, 2013). Such information cannot be obtained
Therefore, a forward model is generally not applicable for boulders de- quantitatively from field observations or inverse models.
posited by historical or pre-historic events whose source models and Forward model may also be useful in establishing criteria for identi-
corresponding flows are unknown. However, if a tsunami source fying tsunami boulders. For example, if the source region of the boulder
model is estimated from other evidence such as historical documents, and the local topography are known, a forward model can be used to
a forward model is applicable to check the validity of the source discriminate tsunami boulders from boulders transported by other
model using field observations and modeling of boulder transport high-energy events. The major difference between a tsunami and
(Figs. 7 and 8, Imamura et al., 2008; Goto et al., 2010a). storm wave is the wave period rather than the wave height (Barbano
et al., 2010; Goto et al., 2010c). Using this difference, it is possible to
5.2.3. Inverse model evaluate how long of a wave period (plus wave height) is required to
Inverse models use simple hydrodynamic equations to calculate the explain the transport distance from the original to the present positions.
minimum flow depth/flow velocity to initiate boulder transport by tsu-
namis or storms (Nott, 1997, 2003; Noormets et al., 2004; Benner et al., 5.3.2. Improvement of the model
2010; Nandasena et al., 2011b) or to calculate an inundation area Many difficulties remain in achieving more accurate calculation of
(Pignatelli et al., 2009, 2011). Lorang (2011) and Weiss (2012) pro- both forward and inverse models because of the uncertainties of boul-
posed theoretical models to account for the effect of differences in tsu- ders and hydrodynamic force acting on the boulders (e.g., Imamura
nami wavelength or energy on boulder transport. Goff et al. (2010) et al., 2008; Nandasena et al., 2011a,b; Nandasena and Tanaka, 2013).
theoretically considered whether boulders can be transported by sus- For example, accurate estimation of the size, shape, densities and poros-
pension load. Hansom et al. (2008) and Kogure and Matsukura (2010) ities of boulders is a critical problem in determining the input parame-
conducted modeling and stability analysis for cliff-top boulders involv- ters for boulders. The terrestrial laser scanner technique that was used
ing wave pressure by tsunamis or storm waves. by Scicchitano et al. (2012) may useful to better estimate the size and
Inverse models are used by geologists/geomorphologists to discrim- weight of boulders.
inate between tsunami and storm transport of boulders (Nandasena Kain et al. (2012) argued that water alone was not able to initiate the
and Tanaka, 2013) because the only model inputs are properties of boul- transport of the boulders, and the role of sediment in increasing the
ders (e.g. size, density, porosity and shape of boulders; Spiske et al., density of water must be assessed because the greater fluid density
Fig. 7. Example of the forward model calculation adopted for a boulder at the Miyara Bay of Ishigaki Island that was dated by Kato and Kimura (1983) as AD1771, Meiwa tsunami origin. We
assumed that the original position of the boulder was around the reef edge and investigated their transport process by assuming a fault model proposed by Imamura et al. (2008). The path
of the No. 8 boulder, which was deposited very close to the observed location, is shown as a red line (courtesy: K. Okada).
increased the forces on boulders. They suggested that the model needs variety of sedimentary and remote-sensing data. This data provides a
to take into account smaller sediment and eventually move towards a great opportunity to advance numerical models. The data consists of
Bingham flow concept in the case of tsunami waves. Collision of boul- the following:
ders is not considered in the forward models, but it is also an important
1. Nearshore and offshore tsunami records obtained from GPS tide sta-
factor for the transport distance of boulders, according to the water tank
tions (e.g., DART and NOWPHAS)
experiment (Imamura et al., 2008; Nandasena and Tanaka, 2013).
2. More than 5000 measurements of tsunami and run-up heights (Mori
The choice of parameter values, such as coefficients, is critically im-
et al., 2012)
portant for accurate calculations. For example, the choice of the drag co-
3. Detailed maps of inundated area identified based on satellite imager-
efficient is very important for both the forward and inverse models
ies, aerial photographs and field surveys (e.g., Haraguchi and
because drag force is the largest force acting on the boulder. Neverthe-
Iwamatsu, 2011a,b)
less, many researchers use different values ranging from 1.05 to 2.0
4. Sedimentary data from tsunami deposits, ranging from clays to huge
(e.g., Nott, 2003; Noormets et al., 2004; Imamura et al., 2008;
boulders, gathered from coastal plains and narrow valleys at more
Nandasena et al., 2011a,b). As Imamura et al. (2008) noted, the Reyn-
than 1000 pits (e.g., Goto et al., 2011a, 2012a,b; in press)
olds number of tsunamis in nature is very high, on the order of 107,
5. Airborne video footage that captured tsunami inundation process
which is not reproduced in water tank experiments used to estimate
(e.g., Goto et al., 2011a; Hayashi and Koshimura, 2012; Tappin
the drag coefficient. Therefore, without rigorous determination of each
et al., 2012)
parameter when specifically reproducing tsunami-like currents with a
6. Geodetic data from GPS stations (Ozawa et al., 2011)
very high Reynolds number, it should be understood that both the for-
ward and inverse models have more or less uncertainty. Each type of data is invaluable for forward and inverse modeling of
In contrast, the pre-transport angle of paleo-event boulders is usual- tsunami propagation, inundation and sediment transport. One remark-
ly an uncertain parameter, and is indeterminable from the field observa- able feature of the massive dataset is the availability of pre- and post-
tions in most cases. Nevertheless, it is important for the calculation of tsunami satellite images and aerial photographs, as well as high-
the maximum transport distance (Imamura et al., 2008; Nandasena resolution digital elevation models (DEMs) obtained by remote sensing
and Tanaka, 2013). Moreover, centimeter-scale micro-topography that technology, such as LIDAR and photogrammetry. The data are now pub-
may not be reflected in the topographic data of the forward model licly available as Fundamental Geospatial Data (Geospatial Information
may be important to stop the boulders. Therefore, it is important to un- Authority of Japan; http://www.gsi.go.jp/kiban/index.html). Grain-size
derstand that there is a limitation to the model, and the modeling re- compositions of sediment from the beach (Matsumoto, 1985) and sea-
sults always have error, which should be reported. bottom (Suzuki and Saito, 1988), micropaleontological data (Matoba,
1976) are also available.
6. Significance of groundbreaking data from the 2011 Tohoku-oki Fig. 9 is an example of a comparison of part of the pre- and post-
tsunami tsunami DEMs of the northern Sendai Plain. The tsunami hit the coast-
line with a 10-m-high wave (Mori et al., 2012) and inundated the
As stated in the previous sections, both the forward and inverse plain 4–5 km (Haraguchi and Iwamatsu, 2011a,b). Sandy to muddy tsu-
models of sandy sediment and boulder transport are difficult to validate nami deposits with a thickness of several tens of centimeters to millime-
because of lack of benchmarks. The 2011 Tohoku-oki earthquake tsuna- ters were left in low-lying rice paddies (e.g. Abe et al., 2012; Goto et al.,
mi offers plenty of seismic, geodetic and tsunami records, as well as a 2012c; Takashimizu et al., 2012). Significant erosion of the sand bar near
Fig. 8. Snapshots of forward modeling at Miyara Bay. Note that boulders (green dots) were transported landward by the first wave (courtesy: K. Okada).
the mouth of the Natori River (Fig. 9) is evident in the post-tsunami bottom of rivers and sea, taking into account the field observations.
data. The general distribution of the areas of erosion and deposition The erosion and deposition by the Tohoku-oki tsunami seems to have
can be resolved based on the difference in the DEMs, after accounting been affected by anthropogenic topographic features (Tappin et al.,
for coseismic subsidence measured by GPS stations (Fig. 10). The differ- 2012), so it is expected that defining the spatial distribution of such fea-
ence in the elevation before and after the 2011 event may include many tures would improve the accuracy of the modeling.
other factors, such as coseismic horizontal displacement, local subsi- Fig. 12 shows an example of the simulated post-tsunami topogra-
dence due to the liquefaction of surficial sediment, seasonal change in phy, as well as the distribution of erosion and deposition, 3 h after the
the beach profile, and anthropogenic land modification, which is a pos- earthquake (modified after Sugawara and Takahashi, 2013). The simu-
sibility given the ~5 years between the pre- and post-tsunami surveys. lation reproduced the massive erosion of the sand bar to the north of
A thorough investigation is needed to exclude effects from these factors. the river mouth (Fig. 12a). Although the sediment transport model as-
Nevertheless, the pre- and post-tsunami DEMs may be useful to validate sumes that the grain size of the sediment is uniform and includes
the modeling result. some empirical parameters such as saturated concentration of
The pre- and post-tsunami satellite images and aerial photographs suspended load, the simulation generally reproduced the erosion of sed-
can be used to identify the land use and initial distribution of movable iment. In addition, the effects from local topographic features and the
sediment. Fig. 11 illustrates the land use of the area, which is modeled initial sediment distribution are reflected in the distribution of erosion
based on the pre-tsunami orthogonal photographs. The extent of and deposition (Fig. 12b).
water bodies, bare ground including beaches, forest, rice paddies, dry Nevertheless, a one-to-one comparison between simulation and ob-
farms, paved surfaces and buildings are identified with a spatial resolu- servation is quite difficult. The thicknesses of the tsunami deposits along
tion that is the same as the DEMs. For the modeling of tsunami inunda- a survey line are in general agreement within a factor of two; however,
tion, the spatial variation of the roughness coefficient is set according to only qualitative comparison is possible so far (Sugawara and Takahashi,
the type of land use. For the modeling of tsunami sediment transport, 2013). The cause of this inconsistency can be explained by various fac-
the source of the sandy sediment is assigned as the beach, forest, and tors that were not considered in this modeling, such as transport of
Fig. 9. Pre-tsunami (a) and post-tsunami (b) digital elevation model of part of the Sendai Plain. The yellow star shows the location of the GPS station where coseismic subsidence of 28 cm
was observed for the 2011 Tohoku-oki earthquake. The thick black solid line in (b) shows the tsunami inundation area, which was identified based on aerial photographs and field obser-
vations (Haraguchi and Iwamatsu, 2011a,b). In this cutout, the pre-tsunami elevation was measured during the winter season between 2005 and 2006, and the post-tsunami topography
was surveyed during late March 2011. The spatial resolutions of both data are 5 m (Modified after Sugawara and Takahashi, 2013).
sediment with mixed grain size (Apotsos et al., 2011c) or a change in Tohoku-oki tsunami, and further progress of the modeling research
tsunami hydrodynamics due to an increasing amount of debris will be achieved through such exploration.
(Ontowirjo et al., 2012). In addition, local variability of tsunami deposits
and a gap between the spatial scale of observation and simulation may 7. Discussion
account for the poor agreement. In this comparison, the observation of
the tsunami deposit was made at microtrenches with a typical width 7.1. Linking forward and inverse models
of tens of centimeters, whereas the spatial resolution of the simulation
is 5 m. Considering the local variability of tsunami deposits, it is not sur- The ultimate goal of the numerical modeling of tsunami sediment
prising that a one-to-one comparison between simulation and observa- transport is to be able to better assess tsunami hazards (Jaffe et al.,
tion becomes a qualitative attempt. Processing of both simulated and 2012). Tsunami hazard is typically quantified by the inundation area,
observed data is needed to close the gap of qualitatively different data run-up height, and flow depth and speed. The question still remains
and to validate the numerical models of sediment transport. There is about how much information about a tsunami can be extracted from
more room to investigate how to utilize the available data on the 2011 its sandy and boulder deposits (e.g., Jaffe et al., 2011, 2012; Nandasena
Fig. 10. Comparison of the pre- and post-tsunami DEMs. The meaning of the yellow star and thick black line is same as Fig. 8. Before making the comparison, the post-tsunami DEM was
corrected using the observed coseismic subsidence of 28 cm (Modified after Sugawara and Takahashi, 2013).
Fig. 11. Example of land use map for the Sendai Plain. Values for the Manning's roughness coefficient (n) are based on Kotani et al. (1998) (Modified after Sugawara and Takahashi, 2013).
Fig. 12. Example of forward modeling of sediment transport by the 2011 Tohoku-oki tsunami (modified after Sugawara and Takahashi, 2013). The solid blue line shows the simulated
inundation area, and the thick black solid line shows the tsunami inundation area identified based on aerial photographs and field observations (Haraguchi and Iwamatsu, 2011a,b).
The hydrodynamic model of the tsunami propagation and inundation (TUNAMI-N2; Goto et al., 1997) is coupled with a sediment transport model (Takahashi et al., 2000, 2008; Gusman
et al., 2012). A uniform grain size of 0.254 mm was applied in the modeling, in reference to the available pre-tsunami data on beach sediment (Matsumoto, 1985). The composite tsunami
source model proposed by Sugino et al. (2013) was used to reproduce the observed coseismic subsidence and offshore tsunami waveform, and the tsunami height and inundation area was
calibrated to fit the corresponding measured data.
and Tanaka, 2013). Inverse modeling of tsunami deposits can provide and boulder tsunami deposits are a promising tool to increase the infor-
information about a tsunami, including an interpretation of the number mation derived from the deposits (Imamura et al., 2008; Li et al., 2012b).
of large waves and an estimation of flow speeds (e.g., Jaffe and The ability to estimate the focal mechanism and magnitude of the earth-
Gelfenbaum, 2007; Jaffe et al., 2011, 2012). Forward models for sandy quake that generated a tsunami from its deposits is desirable both
because most tsunamis are generated by earthquakes and the value of reflecting local geological and geomorphological settings. By using for-
knowing earthquake hazard. However, there are challenges that must ward modeling, we can understand the relationship between varying
be overcome before integrated assessment of earthquake–tsunami haz- hypothesized source characteristics and wave/flow parameters, includ-
ards is possible. ing the number of waves that are effective in the erosion and deposition
For example, recent publications show that tsunami inundation area of sediments. As illustrated by Li et al. (2012a), forward modeling will
and the distribution of sandy deposits does not necessarily reflect all of provide guidance on the best locations, in terms of where the most in-
the earthquake slip and is not directly related to earthquake magnitude. formation can be extracted from deposits, for a geological survey and
The rupture process of the 2011 Tohoku-oki earthquake can be charac- application of inverse models.
terized by two different focal mechanisms (Satake et al., 2013). Very Another advantage of forward modeling is that it is capable of delin-
large slip in the deeper part of the plate interface was followed 3 min eating the time evolution of a tsunami wave/flow field and sediment
later by a huge slip in the shallower part of the plate boundary. Models transport (Apotsos et al., 2011b; Gusman et al., 2012; Li et al., 2012a,
show that the shallow, huge slip generated a short-wavelength, high- b). The analysis of the processes of transport and deposition of sediment
amplitude impulsive wave that hit the coastline of northeast Japan as for the simulated tsunami hydrodynamics will define what information
tsunami higher than 20 m. However, the shallower slip was not found about the tsunami can be recorded in the tsunami deposit. Existing in-
to have contributed to the extensive tsunami inundation on the Sendai verse models are formulated using simplifying assumptions, such as
Plain. The extensive inundation area can be explained only by the very steadiness of the flow, that must be satisfied for the model to work
large slip of the deeper part of the plate interface (Satake et al., 2013). (Jaffe and Gelfenbaum, 2007). Forward modeling will provide a guid-
This implies that the inundation area and the distribution of tsunami de- ance on where and when the assumptions of inverse models are likely
posit do not necessarily reflect the entire slip distribution of the earth- applicable or violated. In addition, such analysis will give us insights
quake. Even if forward modeling explains the distribution of a tsunami into how best to compare forward and inverse models. For example,
deposit, only part of the rupture can be estimated. However, a highly we can investigate how the maximum instantaneous flow speed at a
impulsive wave also impacted the Sendai Bay in the 2011 tsunami specific point calculated by forward modeling compares to the flow
(Satake et al., 2013). The height and flow speed of the tsunami in the speed estimated by inverse modeling, which may be an average over
nearshore might have been increased due to the shallow huge slip and time.
resulting impulsive wave. This would change the sediment transport, A critical step is developing a hybrid model is to link the wave/flow
such as was shown in the numerical experiment by Gusman et al. parameters estimated from the inverse modeling with the wave/flow
(2012). field simulated by the forward modeling of tsunami hydrodynamics.
Enhanced linkage between forward and inverse modeling may be Given that the forward model includes the essential physics of tsunamis
needed to estimate the magnitude information as well as the source and model setups (such as geological/geomorphological settings), we
mechanism of past tsunamis. Developing a hybrid approach, which can compare the simulation with the estimated data. We can further de-
combines forward and inverse sediment transport and hydrodynamic duce a list of candidate source models based on the comparison; howev-
models, may increase the ability to extract tsunami flow speed and er, a lot of candidates would likely come up if the estimated data is
other information from deposits (Jaffe et al., 2012). The advantages available at a small number of location. The data should be collected
and disadvantages of forward and inverse models are different from from as many locations to better constrain the unknown (Gusman
each other; they may compliment and compensate each other. Of et al., 2012; Li et al., 2012a,b). All available geological data should be
course, the output of forward modeling cannot simply be used to bench- considered to discriminate the optimal source, and the forward model-
mark inverse models, and vice versa. Testing one model against another ing of sediment transport may provide constraints on the source. Note
does not necessarily prove anything (Jaffe et al., 2011). The develop- that many of wave/flow parameters calculated by the forward modeling
ment of the hybrid approach is not a straightforward task. Here, a are often sensitive to the bathymetry and topography models, as well as
draft strategy to combine forward and inverse approaches is suggested bottom roughness and vegetation. Small changes in flow speed predict-
to enhance the utility of the numerical modeling of sandy tsunami ed by the forward modeling may result in large changes in the transport
deposits. and deposition of sediment, and uncertainties in the input to forward
The remarkable advantage of the inverse modeling of sandy tsunami modeling are amplified in its outputs, such as the thickness and distri-
deposit is that it is nearly independent of tsunami source, bathymetry/ bution of deposits (Jaffe et al., 2011). Flow speeds estimated by the in-
topography and tsunami hydrodynamic models. For the integrated as- verse model by Jaffe and Gelfenbaum (2007) are sensitive also to the
sessment of earthquake–tsunami hazards, the tsunami source and choice of bottom roughness. Even if the bottom friction is difficult to de-
wave/flow parameters are unknowns to be estimated. Therefore, if the termine, and if translation from shear velocity to flow speed is not
inverse model is successfully validated, its outputs can be used as a sub- strongly constrained, we can use the shear velocity derived from the in-
stitute for measured data. Another advantage of the inverse modeling is verse modeling as input to a forward model. In this way, the greatest
that the effect of inevitable variability in the field data may have limited weakness of the inverse model, the uncertainty in bottom roughness,
consequence on the results (Jaffe et al., 2011, 2012). To acquire a defin- can be neglected. Sensitivity analysis is needed to quantify the uncer-
itive output the sedimentary data should be collected from the right tainties incorporated into the outputs of the modeling, and we should
places and the data should be applied to the model in the correct way. be aware of their limitations and uncertainties when used in the assess-
The disadvantage of the inverse modeling is the opposite of the advan- ment of tsunami hazards. Continued research may reduce the uncer-
tage. The model output does not provide information about the tsunami tainties and increase the reliability of the assessment.
source, as well as the wave/flow parameters in the areas where sedi- The above discussion is also generally applicable to the boulder de-
mentary data is not collected. posits. Moreover, to better identify boulders of tsunami origin, there
One of the advantages of forward modeling is that it is suitable to ex- have been some attempts to use both forward models for wave propa-
plore multiple scenarios of tsunami generation, propagation, inundation gation and/or boulder transport as well as the inverse models
and sediment transport by varying model setup, including different (Barbano et al., 2010; Buckley et al., 2012; Engel and May, 2012).
source models (e.g., Apotsos et al., 2011b, c; Gusman et al., 2012; Li These authors estimated the local wave properties of tsunami and/or
et al., 2012a, b; Cheng and Weiss, 2013). The tsunami source is specified storm waves based on forward modeling and estimated the possible
to run the forward model, and the source mechanism and magnitude maximum current velocities by these extreme wave events. They also
are explicitly linked with the model output. Differences in the source calculated the minimum current velocity necessary to move the boul-
characteristics will appear as changes in the tsunami hydrodynamics ders from their sizes. Then, they concluded that the deposition of
and sediment transport that vary spatially as well as temporally, some boulders can be explained both by tsunami and storm waves,
whereas some inland boulders cannot be explained by the storm wave inevitably differ from the modeling of modern tsunami sediment trans-
generated by the possible maximum storm waves at each area, which port, and criteria to assess the consistency between the simulation and
implies deposition by a tsunami. field data of paleotsunami deposits are especially needed. Geologists
must be familiar with the advantages, limitations and uncertainties in
7.2. Bridging the gap between tsunami geology and numerical modeling the geological data. The advantages, limitations, uncertainties, and fu-
ture challenges for numerical modeling have been addressed in previ-
Knowing the advantages of using numerical modeling of sediment ous chapters. Modelers are familiar with the limitations and
transport, geologists and modelers may have increased their interest uncertainties of numerical models. An important challenge for future
in collaborating on the modeling and geological study of tsunami de- development of this research is for geologists and modelers to exchange
posits. Encouraging recent results include the potential of numerical their understanding of the limitations and uncertainties with each
modeling to estimate flow speeds of paleotsunami events (Witter other.
et al., 2012), to constrain possible seafloor displacement close to the
study area with tsunami deposits (Li et al., 2012b), and estimate the 8. Concluding remarks
source of a large tsunamigenic earthquake, given that sedimentary
data are available from many locations (Gusman et al., 2012). The forward modeling of sandy tsunami deposits can illustrate the
Like the validation of modeling using recent examples, a compre- processes of erosion and deposition of sediment and resulting morpho-
hensive dataset on relevant paleotsunami events is essential to run logical change. Forward models for boulder deposits are useful for un-
the numerical models and to investigate the magnitude information of derstanding boulder transport processes as well as for estimating the
the tsunami. However, such a dataset may include inherent uncer- hypothetical wave source of paleotsunamis. Flow speed and depth of
tainties from many factors. modern and paleotsunami inundations can be estimated using the in-
For example, an important model input that may be poorly verse modeling of sandy and boulder tsunami deposits.
constrained is paleo-topography. In the shallow seas, the behavior of Application of numerical models to paleotsunami deposits is a grand
tsunamis becomes more sensitive to local bathymetry. Onshore topog- challenge because of the limitations and uncertainties in both the model
raphy is an important control for tsunami hydrodynamics and sediment and the geological data. Close collaboration between geologists and
transport. In addition, nearshore and onshore sediment supply strongly modelers will be needed to quantify the magnitude information of
influence the volume of deposition inland. Use of detailed topography paleotsunamis and to assess future tsunami risks.
and geomorphology data of coastal areas before and after a tsunami is es- Existing numerical models do not address gravelly and muddy tsu-
sential to accurately simulate sediment transport (Gusman et al., 2012). nami deposits. Considering the compositional diversity and complexity
This is true and applicable for some of the recent tsunamis, such as the of tsunami deposits and the growing geological interests in these de-
2011 Tohoku-oki event, and true but not applicable at all for posits, future investigations of numerical modeling of gravelly and
paleotsunamis. To reconstruct the paleotopography and geomorphology, muddy sediment is desirable. The difference between mud, sand, gravel
an extensive list of factors is required, including seaward/shoreward mi- and boulders is not only in their size but also in the dominant physical
gration of the coastline, burial/denudation of paleo-ground surface, distri- processes that affect their transport and deposition. An integrated
bution and composition of sediment and vegetation, as well as model applicable to all grain sizes may greatly enhance our understand-
geomorphological change due to the paleotsunamis themselves. Whether ing of tsunami sedimentology and hydrodynamics and improve tsuna-
the reconstruction of paleo-geomorphology is successful or not depends mi hazard assessment from deposits.
on the geological and geomorphological context of the study area. An out-
line can be mapped (e.g. Satake et al., 2008; Namegaya et al., 2010; Acknowledgements
Sugawara et al., 2011); however, reconstruction of the detail does not
seem to be feasible. Moreover, there are no criteria to assess the appropri- We would like to sincerely acknowledge Dr. John T. Wells, the editor,
ateness of the restored map. and two anonymous reviewers for providing their valuable comments
Another important factor to take into account is the possible loss of and suggestions to improve the manuscript. Steve Watt assisted in cre-
information during the post-depositional change of tsunami deposits ation of figures. Part of this research was supported by Grants-in-Aid
(Jaffe et al., 2011, 2012; Szczuciński, 2012). Sandy tsunami deposits from a Grant-in-Aid from JSPS for forward modeling of sediment trans-
are subjected to a reworking by a variety of agents, such as rainfall, port section (D. Sugawara: no. 22241042) and MEXT for boulder trans-
wind action, bioturbation and human activity. For example, in tropical port section (K. Goto: no. 23684041). The Tsunami Hazards, Modeling,
environments the post-depositional soil-forming processes disturb the and the Sedimentary Record project of the US Geological Survey Coastal
internal sedimentary structure of tsunami deposits, such as the vertical and Marine Geology Program supported BEJ's contribution (inverse
variation of grain-size distribution, due to mixing with organic matter modeling of sediment transport) to this research.
from vegetation cover (Szczuciński, 2012). Such processes also may re-
duce the thickness of recognizable tsunami deposits. The effect of post- Appendix A
depositional change may be more severe for inverse modeling of sandy
tsunami deposits, which utilize the grain size variation of the sediment Formulations for calculating the amount of bed and suspended loads
deposited from suspension (Jaffe et al., 2011). Considering the preserva- are either an explicit or implicit function of Shield's parameter. Most of
tion potential of tsunami deposits, the limitations of the numerical the available pick-up rate formulations have been calibrated based on
modeling to extract hydrodynamic information from tsunami deposits flume experiments and field data of ‘moderate’ sedimentological agents,
should be carefully considered (Jaffe et al., 2011; Szczuciński, 2012; such as fluvial, tidal and other normal coastal processes (Apotsos et al.,
Ontowirjo et al., 2012). It is necessary to assess the state of preservation 2011b). Note that the typical flow speed of these agents is on the order
and to identify residual information that infers the hydrodynamic con- of 0.5-1 m/s (Allen, 1984). However, as estimated by video footage from
dition of the tsunami before applying sediment transport modeling to recent tsunamis, the flow speed during a tsunami inundation ranges
paleotsunamis (Jaffe et al., 2011, 2012). from 2–5 m/s (Fritz et al., 2006) and even reaches up to 8 m/s
The numerical modeling of sediment transport by paleotsunamis, at (Hayashi and Koshimura, 2012), and the typical period of a tsunami
least at the present, should accept the uncertainties in the model setup wave is tens of minutes. It is highly likely that the Shield's parameter
and geological data. However, without understanding the limitations during actual tsunami inundation is considerably greater than that of
involved in the uncertainties, incorrect conclusions will be derived. Ap- the moderate agents. Even in the earlier literature, the applicability of
proaches for applying numerical modeling to paleotsunamis will the available formulations of the pick-up rates of bed and suspended
loads has been questioned (Takahashi et al., 2000). At present, the ap- feature (houses, open areas, etc.) classification (e.g., Li et al., 2012b).
propriateness of simple extrapolation of the empirical formulations to The spatial distribution and grain-size composition of sediment also
extreme flow condition by tsunamis is unclear (Apotsos et al., 2011b). need to be specified (Apotsos et al., 2011a; Gusman et al., 2012). Maps
Nevertheless, encouraging results have been reported in the literature; can also be used to specify where, how much, and what type of sedi-
most of studies demonstrated general or even excellent agreements be- ment exists prior to the tsunami. Satellite images and aerial photo-
tween simulation and observation. graphs can be used to make a land use map based on visual
Manning's roughness is commonly used to calculate bottom shear observation of the distribution of water body, vegetation and artificial
stress under an assumption of steady flow, which means the flow objects. Unfortunately, some of the previous studies have suffered
speed and depth does not change in time. However, unsteadiness is an from the uncertainty in the pre-tsunami topography and initial distribu-
important feature of tsunamis. Change in or even reversal of the flow dition of sediment. A simplified assumption of sediment source is intro-
rection may occur depending on the tsunami wave and local topogra- duced to the model setup. The sensitivity of model results to the
phy. In addition, during tsunami inundation on land, a considerable unknown initial conditions can be evaluated by running multiple sce-
lateral gradient of the flow speed may appear near the wave front narios of possible initial sediment distributions.
(e.g., Apotsos et al., 2011b). These variations in the flow condition inher-
ently accompany temporal acceleration and deceleration of the flow. In Appendix D
such situations, the flow is no longer steady. Analytical solutions and
laboratory measurements showed that the shear stress changes its One of the most valuable offshore datasets is from Kesennuma Bay in
sign during the flow deceleration, which means that a phase difference northeast Japan. The 1960 Chilean tsunami hit the bay and caused se-
between the flow speed and bottom shear stress occurs (Liu et al., vere damage. The bathymetry was surveyed before and after the tsuna-
2007). Adityawan and Tanaka (2012) addressed the importance of ac- mi. The volumes of erosion and deposition inside the bay were
counting for the phase difference in the modeling of sediment transport. estimated to be 2,618,000 m3 and 725,000 m3, respectively (Takahashi
They further noted that the bottom shear stress estimated by Manning's et al., 2000). The flow direction and speed inside the bay were estimated
roughness is likely underestimated in deeper water and overestimated using aerial photographs taken during the tsunami, and the tide record
in shallower water. Note that the resistance to the flow may vary de- from a nearby site is available. Therefore, validation of the model using
pending on the land condition. For example, a smooth paved surface the dataset from the Kesennuma Bay was a focus of earlier research
should exert less friction to the flow, whereas dense vegetation on the (Takahashi et al., 2000). The 2010 Chilean and the 2011 Tohoku-oki
surface may provide great resistance against the flow. A variety of stud- earthquake tsunamis also hit the bay, and the bathymetry was mea-
ies has been conducted to determine the appropriate value of bottom sured again before and after these tsunami events (Haraguchi et al.,
roughness to each land condition (summary in Jaffe et al., 2012). Fur- 2012, 2013). These data await further investigation through sediment
thermore, tsunamis likely change the degree of friction on the land sur- transport modeling. Another benchmark came from the Kirinda fishery
face due to the modification of the land condition. For example, harbor in southeast Sri Lanka. The bathymetry change of the harbor was
tsunamis can strip surficial vegetation. Of course, this should be taken measured across the 2004 IOT, just one month before and after the tsu-
into account in future sophisticated numerical models (Apotsos et al., nami (Takahashi et al., 2008). Although the waveform and arrival time
2011a, c). However, like the modeling of tsunami sediment transport, of the tsunami are known for some distant locations, field data that
the modeling of tsunami impacts on the vegetation and change in the identifies the inundation area and tsunami height are available around
bottom roughness is another developing field of research (Imai et al., the harbor (Wijetunge, 2006). Based on the dataset from the harbor,
2009; Yanagisawa et al., 2010). previous studies have investigated the validity of sediment transport
models (Nishihata et al., 2006; Takahashi et al., 2008; Kihara and
Appendix B Matsuyama, 2010) as well as detailed processes of erosion and deposi-
tion (Kihara and Matsuyama, 2010).
Many field observations have demonstrated that topographic undu- A comprehensive dataset for land has been derived from Lhok Nga in
lations (e.g., ridge and swale systems) affect the thickness of sandy on- northwest Sumatra, Indonesia. The inundation area, flow depths and
shore tsunami deposits (e.g., Nishimura and Miyaji, 1995; Gelfenbaum geological data were obtained through post-tsunami field surveys of
et al., 2007). Furthermore, the thickness of tsunami deposits is some- the 2004 IOT (e.g., Jaffe et al., 2006; Moore et al., 2006; Lavigne et al.,
times quite sensitive to the microtopography, including those associat- 2009). The dataset from Lhok Nga has been widely used for previous
ed with natural and artificial structures (e.g., Gelfenbaum and Jaffe, studies of numerical models (Gusman et al., 2012; Li et al., 2012b;
2003; Takashimizu et al., 2012). Forward modeling has also demon- Ontowirjo et al., 2012). Data from Kuala Meurisi, Sumatra for the 2004
strated a significant effect from local topography on the spatial distribu- IOT (Apotsos et al., 2011a,b) and Fagafue Bay in American Samoa for
tion patterns of erosion and deposition and the characteristics of the 2009 Samoa tsunami (Apotsos et al., 2011c) were used also in pre-
tsunami deposits (Gusman et al., 2012; Li et al., 2012b). Even small- vious investigations.
scale features such as walls, roads, trenches or rice paddies may affect tsu-
nami hydrodynamics, sediment transport and deposition (Ontowirjo References
et al., 2012). The typical horizontal resolution of topography data used
in previous studies is on the order of tens of meters. If small-scale features Abe, T., Goto, K., Sugawara, D., 2012. Relationship between the maximum extent of tsuna-
mi sand and the inundation limit of the 2011 Tohoku-oki tsunami on the Sendai
play an important role in sediment transport and deposition, higher reso- Plain. Sedimentary Geology 282, 142–150.
lution topography is needed to resolve these small-scale phenomena. Re- Adityawan, M.B., Tanaka, H., 2012. Bed stress assessment under solitary wave run-up.
cent developments in remote sensing, such as Laser Imaging Detection Earth, Planets and Space 64, 945–954.
Aida, I., 1977. Simulations of large tsunamis occurring in the past off the coast of the
and Ranging (LIDAR; e.g., Fig. 9), as well as in ground surveying using Sanriku District. Bulletin of the Earthquake Research Institute 52, 71–101.
DGPS equipment, allows collection of the needed high-resolution topog- Allen, J.R.L., 1984. Sedimentary structures: their characteristics and physical basis (Unabridged
raphy data. Such data has a horizontal resolution of meters or less and a one-volume edition). Developments in sedimentology, 30. Elsevier (663 pp.).
Apotsos, A., Buckley, M., Gelfenbaum, G., Jaffe, B., Vatvani, D., 2011a. Nearshore tsunami
vertical precision of tens of centimeters or less.
inundation model validation: toward sediment transport applications. Pure and Ap-
plied Geophysics 168, 2097–2119.
Appendix C Apotsos, A., Gelfenbaum, G., Jaffe, B., 2011b. Process-based modeling of tsunami inunda-
tion and sediment transport. Journal of Geophysical Research 116, F01006. http://
dx.doi.org/10.1029/2010JF001797.
Land use maps (e.g., Fig. 11) are used to assign bottom roughness Apotsos, A., Gelfenbaum, G., Jaffe, B., Watt, S., Peck, B., Buckley, M., Stevens, A., 2011c. Tsu-
(Manning's roughness coefficient) to each area according to surface nami inundation and sediment transport in a sediment-limited embayment on
American Samoa. Earth Science Reviews 107, 1–11. http://dx.doi.org/10.1016/ Goto, K., Okada, K., Imamura, F., 2009a. Characteristics and hydrodynamics of boulders
j.earscirev.2010.11.001. transported by storm waves at Kudaka Island, Japan. Marine Geology 262, 14–24.
Araoka, D., Inoue, M., Suzuki, A., Yokoyama, Y., Edwards, R.L., Cheng, H., Matsuzaki, H., Goto, K., Okada, K., Imamura, F., 2009b. Importance of the initial waveform and coastal
Kan, H., Shikazono, N., Kawahata, H., 2010. Historic 1771 Meiwa tsunami confirmed profile for the tsunami transport of boulders. Polish Journal of Environmental Studies
by high-resolution U/Th dating of massive Porites coral boulders at Ishigaki Island 18, 53–61.
in the Ryukyus, Japan. Geochemistry, Geophysics, Geosystems 11, Q06014. Goto, K., Kawana, T., Imamura, F., 2010a. Historical and geological evidence of boulders
Araoka, D., Yokoyama, Y., Suzuki, A., Goto, K., Miyagi, K., Miyazawa, K., Matsuzaki, H., deposited by tsunamis, southern Ryukyu Islands, Japan. Earth-Science Reviews 102,
Kawahata, H., 2014. Tsunami recurrence revealed by Porites coral boulders in the 77–99.
southern Ryukyu Islands, Japan. Geology 41, 919–922. Goto, K., Okada, K., Imamura, F., 2010b. Numerical analysis of boulder transport by the
Ashida, L., Michiue, M., 1972. Study on hydraulic resistance and bed-load transport rate in 2004 Indian Ocean tsunami at Pakarang Cape, Thailand. Marine Geology 268, 97–105.
alluvial streams. Proceedings of the Japanese Society of Civil Engineers 206, 59–69 (in Goto, K., Miyagi, K., Kawamata, H., Imamura, F., 2010c. Discrimination of boulders depos-
Japanese). ited by tsunamis and storm waves at Ishigaki Island, Japan. Marine Geology 269,
Bagnold, R.A., 1966. An approach to the sediment transport problem from general phys- 34–45.
ics. USGS Professional Paper 422-I, 1–37. Goto, K., Chagué-Goff, C., Fujino, S., James Goff, J., Jaffe, B., Nishimura, Y., Richmond, B.,
Barbano, M.S., Pirrotta, C., Gerardi, F., 2010. Large boulders along the south-eastern Ionian Suguwara, D., Szczuciński, W., Tappin, D.R., Witter, R.C., Yulianto, E., 2011a. New in-
coast of Sicily: storm or tsunami deposits? Marine Geology 275, 140–154. sights of tsunami hazard from the 2011 Tohoku-oki event. Marine Geology 290,
Benner, R., Browne, T., Brückner, H., Kelletat, D., Scheffers, A., 2010. Boulder transport by 46–50.
waves: progress in physical modeling. Zeitschrift für Geomorphologie 54, 127–146. Goto, K., Miyagi, K., Kawana, T., Takahashi, J., Imamura, F., 2011b. Emplacement and
Bondevik, S., Løvholt, F., Harbitz, C., Mangerud, J., Dawson, A.G., Svendsen, J.I., 2005. The movement of boulders by known storm waves — field evidence from the Okinawa
Storegga Slide tsunami — comparing field observations with numerical solutions. Ma- Islands, Japan. Marine Geology 283, 66–78.
rine and Petroleum Geology 22, 195–208. Goto, K., Chague-Goff, C., Goff, J., Jaffe, B., 2012a. The future of tsunami research following
Bourgeois, J., 2009. Geologic effects and records of tsunamis. In: Bernard, E.N., Robinson, the 2011 Tohoku-oki event. Sedimentary Geology 282, 1–13.
A.L. (Eds.), The Sea. Tsunamis, 15. Harvard University Press, Cambridge, MA, Goto, K., Sugawara, D., Ikema, S., Miyagi, T., 2012b. Sedimentary processes associated
pp. 55–91. with sand and boulder deposits formed by the 2011 Tohoku-oki tsunami at Sabusawa
Bourgeois, J., MacInnes, B., 2010. Tsunami boulder transport and other dramatic effects of Island, Japan. Sedimentary Geology 282, 188–198.
the 15 November 2006 central Kuril Islands tsunami on the island of Matua. Goto, K., Sugawara, D., Abe, T., Haraguchi, T., Fujino, S., 2012c. Liquefaction as an impor-
Zeitschrift für Geomorphologie 54, 175–195. tant local source of the 2011 Tohoku-oki tsunami deposits at Sendai Plain, Japan. Ge-
Bourgeois, J., Minoura, K., 1997. Paleotsunami studies — contribution to mitigation and ology 40, 887–890.
risk assessment. In: Gusiakov, V.K. (Ed.), Tsunami Mitigation and Risk Assessment, Goto, K., Miyagi, K., Imamura, F., 2013. Localized tsunamigenic earthquakes inferred from
Report of the International Workshop. Petropavlovsk-Kamchatsky, Russia, pp. 1–4 preferential distribution of coastal boulders on the Ryukyu Islands, Japan. Geology 41,
(21–24 August 1996). 1139–1142.
Buckley, M.L., Wei, Y., Jaffe, B.E., Watt, S.G., 2012. Inverse modeling of velocities and in- Goto, K., Hashimoto, K., Sugawara, D., Yanagisawa, H., Abe, T., 2014. Spatial thickness var-
ferred cause of overwash that emplaced inland fields of boulders at Anegada, British iability of the 2011 Tohoku-oki tsunami deposits along the coastline of Sendai Bay.
Virgin Islands. Natural Hazards 63, 133–149. Marine Geology (in press).
Chagué-Goff, C., Schneider, J.-L., Goff, J.R., Dominey-Howes, D., Strotz, L., 2011. Expanding Gusman, A.R., Tanioka, Y., Takahashi, T., 2012. Numerical experiment and a case study of
the proxy toolkit to help identify past events: lessons from the 2004 Indian Ocean sediment transport simulation of the 2004 Indian Ocean tsunami in Lhok Nga, Banda
Tsunami and the 2009 South Pacific Tsunami. Earth-Science Reviews 107, 107–122. Aceh, Indonesia. Earth Planet Space 64, 817–827.
Cheng, W., Weiss, R., 2013. On sediment extent and runup of tsunami waves. Earth and Hall, A.M., 2011. Storm wave currents, boulder movement and shore platform develop-
Planetary Science Letters 362, 305–309. ment: a case study from East Lothian, Scotland. Marine Geology 283, 98–105.
Choowong, M., Murakoshi, N., Hisada, K., Charusiri, P., Charoentitirat, T., Chutakositkanon, Hall, A.M., Hansom, J.D., Williams, D.M., Jarvis, J., 2006. Distribution, geomorphology
V., Jankaew, K., Kanjanapayont, P., Phantuwongraj, S., 2008. 2004 Indian Ocean tsuna- and lithofacies of cliff-top storm deposits: examples from the high energy coasts of
mi inflow and outflow at Phuket, Thailand. Marine Geology 248, 179–192. Scotland and Ireland. Marine Geology 232, 131–155.
Coleman, P.J., 1968. Tsunamis as geological agents. Journal of geological society of Hall, A.M., Hansom, J.D., Williams, D.M., 2010. Wave-emplaced coarse debris and
Australia 15 (2), 267–273. megaclasts in Ireland and Scotland: boulder transport in a high-energy littoral envi-
Dawson, A.G., Shi, S., 2000. Tsunami deposits. Pure and Applied Geophysics 157, 875–897. ronment: a discussion. Journal of Geology 118, 699–704.
Dawson, A.G., Stewart, I., 2007. Tsunami deposits in the geological record. Sedimentary Hansom, J., Barltrop, N., Hall, A.M., 2008. Impact of extreme waves on the Atlantic coasts
Geology 200, 166–183. of the British Isles: modeling the processes of cliff-top erosion and deposition. Marine
Dietrich, W., 1982. Settling velocity of natural particles. Water Resources Research 18, Geology 253, 36–50.
1615–1626. Haraguchi, T., Iwamatsu, A., 2011a. Detailed Maps of the Impacts of the 2011 Japan Tsunami.
Einstein, H.A., 1950. The bed-load functions for sediment transportation in open channel Aomori, Iwate and Miyagi prefectures, 1. Kokon-Shoin Publishers, Tokyo (167 pp.).
flows. USDA, Soil Conservation Service, Technical Bulletin 1026, 1–71. Haraguchi, T., Iwamatsu, A., 2011b. Detailed Maps of the Impacts of the 2011 Japan Tsu-
Engel, M., May, S.M., 2012. Bonaire's boulder fields revisited: evidence for Holocene tsunami. Fukushima, Ibaraki and Chiba prefectures, 2. Kokon-Shoin Publishers, Tokyo
nami impact on the Leeward Antilles. Quaternary Science Reviews 54, 126–141. (97 pp.).
Etienne, S., Paris, R., 2010. Boulder accumulations related to storms on the south coast of Haraguchi, T., Takahashi, T., Hisamatsu, R., Morishita, Y., Sasaki, I., 2012. A field survey of
the Reykjanes Peninsula (Iceland). Geomorphology 114, 55–70. geomorphic change on Kesennuma Bay caused by the 2010 Chilean Tsunami and the
Etienne, S., Buckley, M., Paris, R., Nandasena, A.K., Clark, K., Chagué-Goff, C., Goff, J., 2011 Tohoku Tsunami. Journal of JSCE, Series B2 (Coastal Engineering) 68, 231–235.
Richmond, B., 2011. The use of boulders for characterising past tsunamis: lessons Haraguchi, T., Goto, K., Sato, M., Yoshinaga, Y., Yamaguchi, N., Takahashi, T., 2013. Large
from the 2004 Indian Ocean and 2009 South Pacific tsunamis. Earth-Science Reviews bedform generated by the 2011 Tohoku-oki tsunami at Kesennuma Bay, Japan. Ma-
107, 76–90. rine Geology 335, 200–205.
Fichaut, B., Suanez, S., 2011. Quarrying, transport and deposition of cliff-top storm de- Harbitz, C.B., 1992. Model simulations of tsunamis generated by the Storegga Slides. Ma-
posits during extreme events: Banneg Island, Brittany. Marine Geology 283, 36–55. rine Geology 105, 1–21.
Fritz, H.M., Borrero, J.C., Synolakis, C.E., Yoo, J., 2006. 2004 Indian Ocean tsunami flow ve- Hawkes, A.D., Bird, M., Cowie, S., Grundy-Warr, C., Horton, B.P., Hwai, A.T.S., Law, L.,
locity measurements from survivor videos. Geophysical Research Letters 33, L24605. Macgregor, C., Nott, J., Ong, J.E., Rigg, J., Robinson, R., Tan-Mullins, M., Sa, T.T., Yasin,
http://dx.doi.org/10.1029/2006GL026784. Z., Aik, L.W., 2007. Sediments deposited by the 2004 Indian Ocean Tsunami along
Fujino, S., Naruse, H., Matsumoto, D., Sakakura, N., Suphawajruksakul, A., Jarupongsakul, the Malaysia–Thailand Peninsula. Marine Geology 242, 169–190.
T., 2010. Detailed measurements of thickness and grain size of a widespread onshore Hayashi, S., Koshimura, S., 2012. Measurement of the 2011 Tohoku tsunami flow velocity
tsunami deposit in Phang-nga Province, southwestern Thailand. Island Arc 19 (3), by the aerial video analysis. Journal of JSCE, Series B2 (Coastal Engineering) 68,
389–398. 366–370.
Gelfenbaum, G., Jaffe, B.E., 2003. Erosion and sedimentation from the 17 July 1998 Papua Higman, B., Bourgeois, J., 2008. Deposits of the 1992 Nicaragua tsunami. In: Shiki, T., Tsuji,
New Guinea tsunami. Pure and Applied Geophysics 60, 1969–1999. Y., Yamazaki, T., Minoura, K. (Eds.), Tsunamiites Features and Implications. Elsevier,
Gelfenbaum, G., Vatvani, D., Jaffe, B., Dekker, F., 2007. Tsunami inundation and sediment pp. 81–103.
transport in vicinity of coastal mangrove forest. Coastal Sediments 07 (2), Hori, K., Kuzumoto, R., Hirouchi, D., Umitsu, M., Janjirawuttikul, N., Patanakanog, B., 2007.
1117–1128. Horizontal and vertical variation of 2004 Indian tsunami deposits: an example of two
Goff, J., Dudley, W.C., de Maintenon, M.J., Cain, G., Coney, J.P., 2006. The largest local tsu- transects along the western coast of Thailand. Marine Geology 239, 163–172.
nami in 20th century Hawaii. Marine Geology 226, 65–79. Huntington, K., Bourgeois, J., Gelfenbaum, G., Lynett, P., Jaffe, B., Yeh, H., Weiss, R., 2007.
Goff, J., Weiss, R., Courtney, C., Dominey-Howes, D., 2010. Testing the hypothesis for tsu- Sandy signs of a tsunami's onshore depth and speed. EOS, transactions, American
nami boulder deposition from suspension. Marine Geology 277, 73–77. Geophysical Union 88, 577–584.
Goff, J., Chagué-Goff, C., Nichol, S.L., Jaffe, B., Dominey-Howes, D., 2012. Progress in Imai, K., Harada, K., Watanabe, O., Esashi, T., Shimanuki, N., Yagi, T., Imamura, F., 2009.
palaeotsunami research. Sedimentary Geology 243–244, 70–88. Tsunami hazard mitigation by using coastal forest in a practical field — case study
Goto, C., Ogawa, Y., Shuto, N., Imamura, F., 1997. IUGG/IOC Time Project, Numerical meth- of Iwanuma-Natori coast in Sendai Bay. Journal of JSCE, Series B2 (Coastal Engineer-
od of tsunami simulation with the Leap-Frog scheme. IOC Manuals and ing) 65, 326–330.
GuidesUNESCO, Paris (130 pp.). Imamura, F., Goto, K., Ohkubo, S., 2008. A numerical model for the transport of a boulder
Goto, K., Chavanich, S.A., Imamura, F., Kunthasap, P., Matsui, T., Minoura, K., Sugawara, D., by tsunami. Journal of Geophysical Research 113, C01008.
Yanagisawa, H., 2007. Distribution, origin and transport process of boulders deposit- Jaffe, B.E., Gelfenbaum, G., 2007. A simple model for calculating tsunami flow speed from
ed by the 2004 Indian Ocean tsunami at Pakarang Cape, Thailand. Sedimentary Geol- tsunami deposits. Sedimentary Geology 200, 347–361. http://dx.doi.org/10.1016/
ogy 202, 821–837. j.sedgeo.2007.01.013.
Jaffe, B., Gelfenbaum, G., Rubin, D., Peters, R., Anima, R., Swensson, M., Olcese, D. Bernales Makino, K., 1968. The Meiwa Tsunami at Yaeyama. author, Ishigaki (462 pp. (in
L., Gomez, J., Riega, P., 2003. Tsunami deposits: identification and interpretation of Japanese)).
tsunami deposits from the June 23, 2001 Peru tsunami. Proceedings of the Interna- Mansinha, L., Smylie, D.E., 1971. The displacement fields of inclined faults. Bulletin of the
tional Conference on Coastal Sediments 2003, CD-ROM. World Scientific Publishing Seismological Society of America 61, 1433–1440.
Corp and East Meets West Productions, Corpus Christi, TX, USA (ISBN 981-238- Martin, M.E., Weiss, R., Bourgeois, J., Pinegina, T.K., Houston, H., Titov, V.V., 2008. Combin-
422-7, 13 pp.). ing constraints from tsunami modeling and sedimentology to untangle the 1969
Jaffe, B.E., Borrero, J.C., Prasetya, G.S., Peters, R., McAdoo, B., Gelfenbaum, G., Morton, R., Ozernoi and 1971 Kamchatskii tsunamis. Geophysical Research Letters 35, L01610.
Ruggiero, P., Higman, B., Dengler, L., Hidayat, R., Kingsley, E., Kongko, W., Lukijanto, http://dx.doi.org/10.1029/2007GL032349.
Moore A., Titov, V., Yulianto, E., 2006. Northwest Sumatra and offshore islands field sur- Mastronuzzi, G., Sansò, P., 2004. Large boulder accumulations by extreme waves along the
vey after the December 2004 Indian Ocean Tsunami. Earthquake Spectra 22, 105–135. Adriatic coast of southern Apulia (Italy). Quaternary International 120, 173–184.
Jaffe, B.E., Buckley, M.L., Richmond, B.M., Strotz, L., Etienne, S., Clark, K., Gelfenbaum, G., Matoba, Y., 1976. Recent foraminiferal assemblages off Sendai, northeast Japan. In:
2011. Flow speed estimated by inverse modeling of sandy sediment deposited by Schafer, C.T., Pelletier, B.R. (Eds.), Marine Sediments Special Publication No. 1, First In-
the 29 September 2009 tsunami near Satitoa, east Upolu. Samoa, Earth-Science Re- ternational Symposium on Benthonic Foraminifera of Continental Margins, Part A,
views 107, 23–37. http://dx.doi.org/10.1016/j.earscirev.2011.03.009. Ecology and biology, pp. 205–220.
Jaffe, B.E., Goto, K., Sugawara, D., Richmond, B., Fujino, S., Nishimura, Y., 2012. Flow speed Matsumoto, H., 1985. Beach ridge ranges and the Holocene sea-level fluctuations on allu-
estimated by inverse modeling of sandy tsunami deposits: results from the 11 March vial coastal plains, Northeast Japan. Science Reports Tohoku University, 7th series
2011 tsunami on the coastal plain near the Sendai Airport, Honshu, Japan. Sedimen- (geography) 35, 15–46.
tary Geology 282, 90–109. http://dx.doi.org/10.1016/j.sedgeo.2012.09.002. Matsutomi, H., Okamoto, K., 2010. Inundation flow velocity of tsunami on land. Island Arc
Kain, C.L., Gomez, G., Moghaddam, A.E., 2012. Comment on ‘Reassessment of hydrody- 19, 443–457.
namic equations: minimum flow velocity to initiate boulder transport by high energy Meyer-Peter, E., Muller, R., 1948. Formulas for bed-load transport. Proceedings of 2nd
events (storms, tsunamis), by N.A.K. Nandasena, R. Paris and N. Tanaka [Marine Ge- IAHR Congress, pp. 39–64 (Stockholm).
ology 281, 70–84]. Marine Geology 319–322, 75–76. Middleton, G.V., 1967. Experiments on density of turbidity currents III Deposition of sed-
Kantha, L.H., Clayson, C.A., 2000. Numerical models of oceans and oceanic processes. In- iment. Canadian Journal of Earth Sciences 4, 475–505.
ternational Geophysics Series, 66. Academic Press, San Diego (940 pp.). Minoura, K., Imamura, F., Takahashi, T., Shuto, N., 1997. Sequence of sedimentation pro-
Kato, Y., Kimura, M., 1983. Age and origin of so-called “Tsunami-ishi”, Ishigaki Island, cesses caused by the 1992 Flores tsunami: evidence from Babi Island. Geology 25,
Okinawa prefecture. Journal of Geological Society of Japan 89, 471–474 (in 523–526.
Japanese with English abstr.). Moore, A.L., Nishimura, Y., Gelfenbaum, G., Takanobu, K., Triv, R., 2006. Sedimentary de-
Kato, Y., Akamine, N., Ohori, K., Tamaki, T., Tamura, K., 1991. Movement of limestone posits of the 26 December 2004 tsunami on the northwest coast of Aceh, Indonesia.
blocks by wind waves — An example by typhoon no.21, 1990, at Zanpa Cape, Earth, Planets and Space 58, 253–258.
Okinawa Island, Southwestern Japan. Proceedings of University of the Ryukyus, Moore, A.L., McAdoo, B.G., Ruffman, A., 2007. Landward fining from multiple sources in a
51, pp. 19–33 (in Japanese). sand sheet deposited by the 1929 Grand Banks tsunami, Newfoundland. Sedimentary
Kawana, T., Nakata, K., 1994. Timing of Late Holocene tsunamis originated around the Geology 200, 336–346. http://dx.doi.org/10.1016/j.sedgeo.2007.01.012.
southern Ryukyu Islands, Japan, deduced from coralline tsunami deposits. Journal Moore, A., Goff, J., McAdoo, B.G., Fritz, H.M., Gusman, A., Kalligeris, N., Kalsum, K., Susanto,
of Geography, Japan 103, 352–376 (in Japanese with English abstr.). A., Suteja, D., Synolakis, C.E., 2011. Sedimentary deposits from the 17 July 2006 West-
Kelletat, D., 2008. Comments to [Dawson, A.G. and Stewart, I., 2007. Tsunami deposits in ern Java Tsunami, Indonesia: use of grain size analyses to assess tsunami flow depth,
the geological record. Sedimentary Geology 200, 166–183]. Sedimentary Geology speed, and traction carpet characteristics. Pure and Applied Geophysics 168 (11),
211, 87–91. 1951–1961. http://dx.doi.org/10.1007/.s00024-011-0280-8.
Kelletat, D., Scheffers, S.R., Scheffers, A., 2007. Field signatures of the SE-Asian Mori, N., Takahashi, T., The 2011 Tohoku Earthquake Tsunami Joint Survey Group, 2012.
megatsunami along the west coast of Thailand compared to Holocene paleo- Nationwide survey of the 2011 Tohoku earthquake tsunami. Coastal Engineering
tsunami from the Atlantic region. Pure and Applied Geophysics 164, 413–431. Journal 54, 1–27.
Kihara, N., Matsuyama, M., 2010. Numerical simulations of sediment transport induced by Morton, R.A., Gelfenbaum, G., Jaffe, B.E., 2007. Physical criteria for distinguishing sandy
the 2004 Indian Ocean tsunami near Kirinda port in Sri Lanka. Proceedings of 32nd tsunami and storm deposits using modern examples. Sedimentary Geology 200,
Conference on Coastal Engineering (Shanghai, China, 6 pp.). 184–207. http://dx.doi.org/10.1016/j.sedgeo.2007.01.003.
Kihara, N., Fujii, N., Matsuyama, M., 2012. Three-dimensional sediment transport process Morton, R.A., Richmond, B.M., Jaffe, B.E., Gelfenbaum, G., 2008. Coarse-clast ridge com-
on tsunami-induced topography changes in a harbor. Earth, Planets and Space 64, plexes of the Caribbean: a preliminary basis for distinguishing tsunami and storm-
787–797. wave origins. Journal of Sedimentary Research 78, 624–637.
Kogure, T., Matsukura, Y., 2010. Instability of coral limestone cliffs due to extreme waves. Morton, R.A., Gelfenbaum, G., Buckley, M.L., Richmond, B.M., 2011. Geological effects and
Earth Surface Processes and Landforms 35, 1357–1367. implications of the 2010 tsunami along the central coast of Chile. Sedimentary Geol-
Kon'no, E., 1961. Geological observation of the Sanriku coastal region damaged by the tsu- ogy 242, 34–51.
nami due to the Chile earthquake in 1960. Contributions Institute Geology Paleontol- Nakamura T., Mizutani, N., Yim, S.C., 2009. A three-dimensional coupled fluid–
ogy Tohoku University 52, 1–45 (in Japanese with English abstract). sediment interaction model with bed-load/suspended-load transport for scour
Kortekaas, S., Dawson, A.G., 2007. Distinguishing tsunami and storm deposits: an example analysis around a fixed structure. Journal of Offshore Mechanics and Arctic Engi-
from Martinhal, SW Portugal. Sedimentary Geology 200, 208–221. neering 131, 031104.
Kotani, M., Imamura, F., Shuto, N., 1998. Tsunami run-up simulation and damage estima- Namegaya, et al., 2010. Numerical simulation of the AD 869 Jogan tsunami in
tion by using GIS. Proceedings of the Coastal Engineering JSCE 45, 356–360 (in Ishinomaki and Sendai plains and Ukedo river-mouth lowland. Annual Report
Japanese). of Active Fault and Paleoearthquake Researches 10, 1–21 (in Japanese, with En-
Kuenen, P.H., Menard, H.W., 1952. Turbidity currents, graded and non-graded deposits. glish Abstr.).
Journal of Sedimentary Petrolology 22, 83–96. Nanayama, F., Shigeno, K., 2006. Inflow and outflow facies from the 1993 tsunami in
Lamarche, G., Pelletier, B., Goff, J., 2010. Impact of the 29 September 2009 South Pacific southwest Hokkaido. Sedimentary Geology 187, 139–158.
tsunami on Wallis and Futuna. Marine Geology 271, 297–302. Nandasena, N.A.K., Tanaka, N., 2013. Boulder transport by high energy: numerical model-
Lane, E.W., Kalinske, A.A., 1941. Engineering calculations of suspended sediments. Trans- fitting experimental observations. Ocean Engineering 57, 163–179.
actions AGU 22, 603–607. Nandasena, N.A.K., Paris, R., Tanaka, N., 2011a. Numerical assessment of boulder transport
Lavigne, F., Paris, R., Grancher, D., Wassmer, P., Brunstein, D., Vautier, F., Leone, F., Flohic, by the 2004 Indian ocean tsunami in Lhok Nga, West Banda Aceh (Sumatra,
F., De Coster, B., Gunawan, T., Gomez, Ch., Setiawan, A., Cahyadi, R., Fachrizal, 2009. Indonesia). Computers & Geosciences 37, 1391–1399.
Reconstruction of tsunami inland propagation on December 26, 2004 in Banda Nandasena, N.A.K., Paris, R., Tanaka, N., 2011b. Reassessment of hydrodynamic equations:
Aceh, Indonesia, through field investigations. Pure and Applied Geophysics 166, minimum flow velocity to initiate boulder transport by high energy events (storms,
259–281. tsunamis). Marine Geology 281, 70–84.
Leliavsky, S., 1955. An introduction to fluvial hydraulics. Constable, London (257 pp.). Nandasena, N.A.K., Tanaka, N., Sasaki, Y., Osada, M., 2013. Boulder transport by the 2011
Li, L., Huang, Z., Qui, Q., Natawidjaja, D.H., Sieh, K., 2012a. Tsunami-induced coastal Great East Japan tsunami: comprehensive field observations and whither model pre-
change: scenario studies for Painan, West Sumatra, Indonesia. Earth, Planets and dictions? Marine Geology 346, 292–309.
Space 64, 799–816. Nishihata, T., Tajima, Y., Moriya, Y., Sekimoto, T., 2006. Topography change due to the Dec
Li, L., Qiu, Q., Huang, Z., 2012b. Numerical modeling of the morphological change in Lhok 2004 Indian Ocean Tsunami — field and numerical study at Kirinda port, Sri Lanka.
Nga, west Banda Aceh, during the 2004 Indian Ocean tsunami: understanding tsuna- Proceedings of 30th International Conference on Coastal Engineering, ASCE,
mi deposits using a forward modeling method. Natural Hazards. http://dx.doi.org/ pp. 1456–1468.
10.1007/s/11069-012-0325-z. Nishimura, Y., Miyaji, N., 1995. Tsunami deposits from the 1993 Southwest Hokkaido
Liu, P.L.-F., Park, Y.S., Cowen, E.A., 2007. Boundary layer flow and bed shear stress under a Earthquake and the 1640 Hokkaido Komagatake Eruption Northern Japan. Pure and
solitary wave. Journal of Fluid Mechanics 574, 449–463. Applied Geophysics 144, 719–733.
Lorang, M.S., 2011. A wave-competence approach to distinguish between boulder Noji, M., Imamura, F., Shuto, N., 1993. Numerical simulation of movement of large rocks
and megaclast deposits due to storm waves versus tsunamis. Marine Geology transported by tsunamis. Procedings of the IUGG/IOC International Tsunami Sympo-
283, 90–97. sium, pp. 189–197.
MacInnes, B.T., Bourgeois, J., Pinegina, T.K., Kravchunovskaya, E.A., 2009. Tsunami geo- Noormets, R., Crook, K.A.W., Felton, E.A., 2002. Sedimentology of rocky shorelines: 2:
morphology: erosion and deposition from the 15 November 2006 Kuril Island tsuna- Shoreline megaclasts on the north shore of Oahu, Hawaii—origins and history. Sedi-
mi. Geology 37, 995–998. mentary Geology 150, 31–45.
MacInnes, B.T., Weiss, R., Bourgeois, J., Pinegina, T.K., 2010. Slip distribution of the 1952 Noormets, R., Crook, K.A.W., Felton, E.A., 2004. Sedimentology of rocky shorelines: 3. Hy-
Kamchatka great earthquake based on near-field tsunami deposits and historical re- drodynamics of megaclast emplacement and transport on a shore platform, Oahu,
cords. Bulletin of Seismological Society of America 100 (4), 1695–1709. Hawaii. Sedimentary Geology 172, 41–65.
Nott, J., 1997. Extremely high-energy wave deposits inside the Great Barrier Reef, Shi, S., Dawson, A.G., Smith, D.E., 1995. Coastal sedimentation associated with the Decem-
Australia: determining the cause-tsunami or tropical cyclone. Marine Geology 141, ber 12th, 1992 tsunami in Flores, Indonesia. Pure and Applied Geophysics 144,
193–207. 525–536.
Nott, J., 2003. Waves, coastal boulder deposits and the importance of the pre-transport Shiki, T., Tachibana, T., Fujiwara, O., Goto, K., Nanayama, F., Yamazaki, T., 2008.
setting. Earth and Planetary Science Letters 210, 269–276. Characteristic features of tsunamiites. In: Shiki, T., Yamazaki, T., Minoura, K.
Nott, J., 2004. The tsunami hypothesis — comparisons of the field evidence against the ef- (Eds.), Tsunamiites — features and implications. Elsevier, Amsterdam,
fects, on the Western Australian coast, of some of the most powerful storms on Earth. Netherlands, p. 411p.
Marine Geology 208, 1–12. Smith, D.E., Foster, I.D.L., Long, D., Shi, S., 2007. Reconstructing the pattern and depth of
Okada, Y., 1985. Surface deformation due to shear and tensile faults in halfspace. Bulletin flow onshore in a palaeotsunami from associated deposits. Sedimentary Geology
of the Seismological Society of America 75, 1135–1154. 200, 362–371. http://dx.doi.org/10.1016/j.sedgeo.2007.01.014.
Onda, M., 1999. The distribution of large reef blocks and the effect of geomorphic devel- Soulsby, R., 1997. Dynamics of marine sands. Thomas Telford Publications, London (270
opment in Kudaka Island, the Ryukyu Islands, Japan. Geographical Review of Japan pp. ISBN 0 7277 2584 X).
72A–11, 746–762 (in Japanese). Soulsby, R.L., Smith, D.E., Ruffman, A., 2007. Reconstructing tsunami run-up from sedi-
Ontowirjo, B., Paris, R., Mano, A., 2012. Modeling of coastal erosion and sediment deposi- mentary characteristics — a simple mathematical model. Coastal Sediments 07 (2),
tion during the 2004 Indian Ocean tsunami in Lhok Nga, Sumatra Indonesia. Natural 1075–1088.
Hazards. http://dx.doi.org/10.1007/s11069-012-0455-3. Spiske, M., Bahlburg, H., 2011. A quasi-experimental setting of coarse clast transport by
Ozawa, S., Nishimura, T., Suito, H., Kobayashi, T., Tobita, M., Imakiire, T., 2011. Coseismic the 2010 Chile tsunami (Bucalemu, Central Chile). Marine Geology 289, 72–85.
and postseismic slip of the 2011 magnitude-9 Tohoku-Oki earthquake. Nature 475, Spiske, M., Bšršcz, Z., Bahlburg, H., 2008. The role of porosity in discriminating be-
373–376. tween tsunami and hurricane emplacement of boulders — a case study from
Paris, R., Wassmer, P., Sartohadi, J., Lavigne, F., Barthomeuf, B., Desgages, E., Grancher, D., the Lesser Antilles, southern Caribbean. Earth and Planetary Science Letters
Baumert, P., Vautier, F., Brunstein, D., Gomez, C., 2009. Tsunamis as geomorphic crises — 268 (3–4), 384–396.
lessons from the December 26, 2004 tsunami in Lhok Nga, West Banda Aceh (Sumatra, Spiske, M., Weiss, R., Bahlburg, H., Roskosch, J., Amijaya, H., 2010. The TsuSedMod inver-
Indonesia). Geomorphology 104, 59–72. sion model applied to the deposits of the 2004 Sumatra and 2006 Java tsunami and
Paris, R., Fournier, J., Poizot, E., Etienne, S., Morin, J., Lavigne, F., Wassmer, P., 2010. implications for estimating flow parameters of paleo-tsunami. Sedimentary Geology
Boulder and fine sediment transport and deposition by the 2004 tsunami in Lhok 224, 29–37. http://dx.doi.org/10.1016/j.sedgeo.2009.12.005.
Nga (western Banda Aceh, Sumatra, Indonesia) — a coupled offshore–onshore Stephenson, W.J., Naylor, L.A., 2011. Geological controls on boulder production in a rock
model. Marine Geology 268, 43–54. coast setting: insights from South Wales, UK. Marine Geology 283, 12–24.
Paris, R., Naylor, L., Stephenson, W.J., 2011. Boulders as a signature of storms on rock Suanez, S., Fichaut, B., Magne, R., 2009. Cliff-top storm deposits on Banneg Island, Brittany,
coasts. Marine Geology 283, 1–11. France: effects of giant waves in the Eastern Atlantic Ocean. Sedimentary Geology
Peters, R., Jaffe, B.E., 2010. Identification of tsunami deposits in the geologic record: devel- 220, 12–28.
oping criteria using recent tsunami deposits. US Geological Survey Open-File Report Sugawara, D., Goto, K., 2012. Numerical modeling of the 2011 Tohoku-oki tsunami in
2010–1239 (39 pp. [URL: http://pubs.usgs.gov/of/2010/1239/]). the offshore and onshore of Sendai Plain, Japan. Sedimentary Geology 282,
Peters, R., Jaffe, B., Gelfenbaum, G., 2007. Distribution and sedimentary characteristics of 110–123.
tsunami deposits along the Cascadia margin of western North America. Sedimentary Sugawara, D., Takahashi, T., 2013. Numerical simulation of coastal sediment transport by
Geology 200, 372–386. the 2011 Tohoku-oki earthquake tsunami. In: Kontar, Y.A., Santiago-Fandino, V.,
Pignatelli, C., Sansò, P., Mastronuzzi, G., 2009. Evaluation of tsunami flooding using geo- Takahashi, T. (Eds.), Tsunami Events and Lessons Learned, Advances in Natural and
morphologic evidence. Marine Geology 260, 6–18. Technological Hazards Research. Springer, pp. 99–112.
Pignatelli, C., Ferilli, S., Capolongo, D., Marsico, A., Milella, M., Pennetta, L., Piscitelli, A., Sugawara, D., Imamura, F., Matsumoto, H., Goto, K., Minoura, K., 2011. Reconstruction of
Mastronuzzi, G., 2011. Morphological evidences and computer science techniques the AD 869 Jogan earthquake induced tsunami by using the geological data. Journal
in order to evaluate tsunami inundation limit. Italian Journal of Remote Sensing 42, of the Japan Society Nature Disasters Science 29 (4), 501–516 (in Japanese, with En-
129–142. glish Abstr.).
Ranasinghe, D.P., Goto, K., Takahashi, J., Wijetunge, J.J., Nishihata, T., Imamura, F., 2013. Sugino, H., Wu, C., Korenaga, M., Nemoto, M., Iwabuchi, Y., Ebisawa, K., 2013. Analysis and
Numerical assessment of bathymetric changes caused by the 2004 Indian Ocean tsu- verification of the 2011 Tohoku Earthquake Tsunami at Nuclear Power Plant Sites.
nami at Kirinda fishery harbor, Sri Lanka. Coastal Engineering 81, 67–81. The Journal of Japan Association for Earthquake Engineering 13, 2–21.
Razzhigaeva, N.G., Ganzei, L.A., Grebennikova, T.A., Ivanova, E.D., Kaistrenko, V.M., 2006. Suzuki, T., Saito, Y., 1988. Heavy mineral composition and provenance of marine sedi-
Sedimentation particularities during the tsunami of December 26, 2004 in Northern ments in Sendai Bay, northern Honshu Japan. Bulletin. Geological Survey of Japan
Indonesia: Simelue Island and the Medan Coast of Sumatra Island. Oceanology 46, 39, 643–660.
875–890. Suzuki, A., Yokoyama, Y., Kan, H., Minoshima, K., Matsuzaki, H., Hamanaka, N., Kawahata,
Ribberink, J.S., 1998. Bed-load transport for steady flows and unsteady oscillatory flows. H., 2008. Identification of 1771 Meiwa Tsunami deposits using a combination of ra-
Coastal Engineering 34, 59–82. diocarbon dating and oxygen isotope microprofiling of emerged massive Porites
Richmond, B., Buckley, M., Etienne, S., Strotz, L., Chagué-Goff, C., Clark, K., Goff, J., McAdoo, boulders. Quaternary Geochronology 3, 226–234.
B., 2011a. Deposits, flow characteristics, and landscape change resulting from the Switzer, A.D., Burston, J.M., 2010. Competing mechanisms for boulder deposition on the
September 2009 South Pacific Tsunami in the Samoan Islands. Earth-Science Reviews southeast Australian coast. Geomorphology 114, 42–54.
107, 38–51. Szczuciński, W., 2012. The post-depositional changes of the onshore 2004 tsunami de-
Richmond, B.M., Watt, S., Buckley, M., Jaffe, B.E., Gelfenbaum, G., Morton, R.A., 2011b. Re- posits on the Andaman Sea coast of Thailand. Natural Hazards 60, 115–133.
cent storm and tsunami coarse-clast deposit characteristics, southeast Hawaii. Ma- Szczuciński, W., Chaimanee, N., Niedzielski, P., Rachlewicz, G., Tepsuwan, T., Lorenc, S.,
rine Geology 283, 79–89. Siepak, J., 2006. Environmental and geological impacts of the 26 December 2004 tsu-
Rinehart, M.A., 1991, Sedimentological analysis of postulated tsunami-generated deposits nami in coastal zone of Thailand—overview of short and long-term effects. Polish
from Cascadia great-subduction earthquakes along southern coastal Washington, un- Journal of Environmental Studies 15, 793–810.
published, University of Washington Department of Geological Sciences M.S. Project, Takahashi, T., Imamura, F., Shuto, N., 1993. Numerical simulation of topography change
75 pp. due to tsunamis. Proceedings of the IUGG/IOC International Tsunami Symposium,
Satake, K., Atwater, B.F., 2007. Long-term perspectives on giant earthquakes and tsunamis pp. 243–255.
at subduction zone. Annual Review of Earth and Planetary Sciences 35, 349–374. Takahashi, T., Shuto, N., Imamura, F., Asai, D., 2000. Modeling sediment transport due to
Satake, K., Wang, K., Atwater, B.F., 2003. Fault slip and seismic moment of the 1700 tsunamis with exchange rate between bedload layer and suspended load layer. Pro-
Cascadia earthquake inferred from Japanese tsunami descriptions. Journal of Geo- ceedings of the International Conference Coastal Engineering 2000, 1508–1519.
physical Research 108 (B11), 2535. http://dx.doi.org/10.1029/2003JB002521. Takahashi, J., Goto, K., Oie, T., Yanagisawa, H., Imamura, F., 2008. Inundation and topo-
Satake, K., Namegaya, Y., Yamaki, S., 2008. Numerical simulation of the AD 869 Jogan tsu- graphic change due to the 2004 Indian Ocean Tsunami at the Kirinda port, Sri
nami in Ishinomaki and Sendai plains. Annual Report of Active Fault and Lanka. Annual Journal of Coastal Engineering, JSCE 55, 251–255.
Paleoearthquake Researches 8, 71–89 (in Japanese, with English Abstr.). Takashimizu, Y., Urabe, A., Suzuki, K., Sato, Y., 2012. Deposition by the 2011 Tohoku-oki
Satake, K., Fujii, Y., Harada, T., Namegaya, Y., 2013. Time and space distribution of tsunami on coastal lowland controlled by beach ridges near Sendai Japan. Sedimen-
coseismic slip of the 2011 Tohoku Earthquake as inferred from tsunami wave- tary Geology 282, 124–141.
form data. Bulletin of the Seismological Society of America 103 (no. 2B), Tappin, D.R., 2007. Sedimentary features of tsunami deposits — their origin, recognition
1473–1492. and discrimination: an introduction. Sedimentary Geology 200, 151–154.
Scheffers, A., Kelletat, D., 2003. Sedimentologic and geomorphologic tsunami imprints Tappin, D.R., Evans, H.M., Jordan, C.J., Richmond, B., Sugawara, D., Goto, K., 2012. Coastal
worldwide—a review. Earth-Science Reviews 63, 83–92. changes in the Sendai area from the impact of the 2011 Tōhoku-oki tsunami: inter-
Scheffers, A., Scheffers, S., 2006. Documentation of the impact of hurricane Ivan on the pretations of time series satellite images, helicopter-borne video footage and field ob-
coastline of Bonaire (Netherlands Antilles). Journal of Coastal Research 22, servations. Sedimentary Geology 282, 151–174.
1437–1450. Titov, V.V., Jaffe, B.E., Gonzalez, F.I., Gelfenbaum, G., 2001. Re-evaluating source mecha-
Scheffers, A., Scheffers, S., Kelletat, D., 2005. Paleo-tsunami relics on the southern and cen- nisms for the 1998 Papua New Guinea tsunami using revised slump estimates and
tral Antillean Island Arc. Journal of Coastal Research 21, 263–273. sedimentation modeling. Proceedings of the International Tsunami Symposium,
Scicchitano, G., Pignatelli, C., Spampinato, C.R., Piscitelli, A., Milella, M., Monaco, C., pp. 389–396.
Mastronuzzi, G., 2012. Terrestrial Laser Scanner techniques in the assessment of tsu- Titov, V.V., Gonzalez, F.I., Bernard, E.N., Eble, M.C., Mofjeld, H.O., Newman, J.C., Venturato,
nami impact on the Maddalena peninsula (south-eastern Sicily, Italy). Earth, Planets A.J., 2005. Real-time tsunami forecasting: challenges and solutions. Natural Hazards
and Space 64, 889–903. 35, 41–58.
Shepard, F.P., Macdonald, G.A., Cox, D.C., 1950. The tsunami of April 1, 1946 [Hawaii]. Van Rijn, L.C., 1984a. Sediment transport, Part I: Bed load transport. Journal of Hydraulic
Scripps Institution of Oceanography Bulletin, 5. California University, pp. 391–528. Engineering 110, 1431–1456.
Van Rijn, L.C., 1984b. Sediment transport, Part II: Suspended load transport. Journal of Hy- Williams, D.M., Hall, A.M., 2004. Cliff-top megaclast deposit of Ireland, a record of extreme
draulic Engineering 110, 1613–1641. waves in the North Atlantic: storms or tsunamis? Marine Geology 206, 101–117.
Van Rijn, L.C., 1993. Principles of Sediment Transport in Rivers, Estuaries and Coastal Witter, R.C., Jaffe, B.E., Zhang, Y., Priest, G., 2012. Reconstruction hydrodynamic flow param-
Areas. Aqua Publications, Amsterdam. eters of the 1700 tsunami at Cannon Beach, Oregon, USA. Natural Hazards 33, 233–240.
Van Rijn, L.C., 2007a. Unified view of sediment transport by currents and waves. I: Initia- http://dx.doi.org/10.1007/s11069-011-9912-7 (plus electronic supplements).
tion of motion, bed roughness, and bed-load transport. Journal of Hydraulic Engineer- Yanagisawa, H., Miyagi, T., Baba, S., 2010. Tsunami mitigation effect of mangrove forest in
ing 133, 649–667. 2009 Samoa earthquake tsunami. Journal of JSCE, Series B2 (Coastal Engineering) 66,
Van Rijn, L.C., 2007b. Unified view of sediment transport by currents and waves II: 251–255.
Suspended transport. Journal of Hydraulic Engineering 133, 668–689. Yawsangratt, S., Szczuciński, W., Chaimanee, N., Jagodziński, R., Lorenc, S., Chatprasert, S.,
Van Rijn, L.C., Walstra, D.J.R., van Ormondt, M., 2004. Description of TRANSPOR 2004 Saisuttichai, D., Tepsuwan, T., 2009. Depositional effects of 2004 tsunami and hypo-
(TR2004) and implementation in DELFT3D-online, Rep. Z3748. Delft Hydraul, Delft, thetical paleotsunami near Thap Lamu Navy Base in Phang Nga Province, Thailand.
Netherlands. Polish Journal of Environmental Studies 18, 17–23.
Wei, Y., Bernard, E., Tang, L., Weiss, R., Titov, V., Moore, C., Spillane, H., Kanoglu, U., 2008. Yoshii, T., Ikeno, M., Matsuyama, M., Fujii, N., 2010. Pick-up rate of suspended sand
Real-time experimental forecast of the Peruvian tsunami of August 2007 for U.S. due to tsunami. Proceedings of the International Conference Coastal Engineering
coastlines. Geophysical Research Letters 35, L04609. 2010, 15p.
Weiss, R., 2012. The mystery of boulders moved by tsunamis and storms. Marine Geology Young, R.W., Bryant, E.A., 1992. Catastrophic wave erosion on the southeastern coast of
295–298, 28–33. Australia: impact of the Lanai tsunamis ca. 105 ka? Geology 20, 199–202.
Weiss, R., Bourgeois, J., 2012. Understanding Sediments — Reducing Tsunami Risk. Science Young, R.W., Bryant, E.A., Price, D.M., 1996. Catastrophic wave (tsunami?) transport of
336, 1117–1118. boulders in southern New South Wales, Australia. Zeitschrift für Geomorphologie
Wijetunge, J.J., 2006. Tsunami on 26 December 2004: spatial distribution of tsunami 40, 191–207.
height and the extent of inundation in Sri Lanka. Science of Tsunami Hazards 24,
225–239.

Sugawara 2014

Uploaded by

Copyright:

Available Formats

Sugawara 2014

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sugawara 2014

Uploaded by

Copyright:

Available Formats

Marine Geology 352 (2014) 295–320

Contents lists available at ScienceDirect

Numerical models of tsunami sediment transport — Current

water column, and ﬂow speed is calculated as a depth-averaged value. If

Ashida and Michiue (1972), Takahashi et al. (2000)

Takahashi et al. (2000), Takahashi et al. (2008),

Takahashi et al. (2000), Yoshii et al. (2010)

Yoshii et al. (2010), Gusman et al. (2012)

Kesen-numa, Lhok Nga, Kirinda

TUNAMI-N2 and others

Sand, single grain-size

Kihara and Matsuyama (2010)

Sand, single grain-size

van Rijn (1984b)

Ontowirjo et al. (2012)

Sand, single grain-size

van Rijn (1984b)

van Rijn (1993), Soulsby (1997)

Sand, mixed grain-size

van Rijn (1993)

real-time tsunami forecasting systems is a good example of an efﬁcient

practical application of forward modeling (e.g., Titov et al., 2005; Wei

modeling has investigated the heights of historical tsunamis

The development of forward numerical models for the transport of

sediment-related effects for modern tsunamis and advancing hazard as-

sessment. Early modeling investigated the impacts of tsunami sediment

transport in coastal areas, such as erosion and deposition of ports and

signiﬁcance and potential of forward modeling to explore paleotsunami

2012b; Ontowirjo et al., 2012; Ranasinghe et al., 2013). To date, research

on forward modeling mostly focuses on deposition and processes of

sediment transport in recent tsunamis.

Table 1 summarizes the existing forward numerical models that

These previous studies include investigations of three modern tsunami

tsunamis), as well as ﬂume experiments (idealized topography) and a

Location Event Morphological data Waveform Inundation Tsunami Flow References

Sea Land Sea Land

Kesennuma 1960 ✓ ✓ ✓ ✓ Takahashi et al. (2000)

Approach Tsunami characteristics estimated Inputs Assumptions References

4.1.2. Particle settling model

4.1.3. Settling column model

Rearranging Eq. (1) gives ws T

You might also like