Pure Appl. Geophys.

Ó 2018 Springer International Publishing AG, part of Springer Nature

https://doi.org/10.1007/s00024-018-1804-2 Pure and Applied Geophysics

Implications on 1 + 1 D Tsunami Runup Modeling due to Time Features of the Earthquake

Source
M. FUENTES,1 S. RIQUELME,2 J. RUIZ,1 and J. CAMPOS1

Abstract—The time characteristics of the seismic source are propagation velocities are by far slower than the
usually neglected in tsunami modeling, due to the difference in the
time scale of both processes. Nonetheless, there are just a few
rupture (Kajiura 1981). Kajiura (1981) demonstrated
analytical studies that intended to explain separately the role of the that there was equivalence between generating a
rise time and the rupture velocity. In this work, we extend an tsunami from a static seafloor displacement or from a
analytical 1 ? 1 D solution for the shoreline motion time series,
dynamic one, because of the much lower velocity of
from the static case to the kinematic case, by including both rise
time and rupture velocity. Our results show that the static case the tsunami propagating through the source. Since
corresponds to a limit case of null rise time and infinite rupture then, this term has not been a subject of importance in
velocity. Both parameters contribute in shifting the arrival time, but tsunami analytical modeling.
maximum runup may be affected by very slow ruptures and long
rise time. Parametric analysis reveals that runup is strictly The evidence of a variety of slow earthquakes
decreasing with the rise time while is highly amplified in a certain (Ide et al. 2007) suggests at least examining and
range of slow rupture velocities. For even lower rupture velocities, testing this hypothesis. The 2004 Sumatra–Andaman
the tsunami excitation vanishes and for larger, quicker approaches
to the instantaneous case.
megathrust earthquake has been studied by many
authors investigating the relation between earthquake
Key words: Tsunami, seismology, runup. and tsunami generation (Ammon et al. 2005; Lay
et al. 2005; Stein and Okal 2005). These studies
showed that there is a slow component on the rupture
towards the north, according to Lay et al. (2005). The
1. Introduction main excitation of the tsunami was located around
500 km from the epicenter source occurring within
The study of tsunamis from analytical approaches the first 500 s. Then, the rupture shows a slow com-
has been treated for decades (e.g. Kajiura 1970; ponent in the next 700–800 km and the Bengal bay
Carrier and Greenspan 1958; Synolakis 1987; Kâno- brings special attention because there are large runup
ğlu 2004; Madsen and Schaffer 2010; Fuentes et al. heights, but no high-frequency earthquake was radi-
2013; Fuentes 2017). However, just a few analytic ated, and buildings did not show signs of damage
studies involve explicit time characteristics of the (Lay et al. 2005). This slow component of the rupture
generation process with applications to tsunamis may play an important role in the tsunami generation.
triggered by earthquakes (Hammack 1973; Dutykh Effects of tsunami amplification along the Japa-
and Dias 2007; Todorovska and Trifunac 2001; nese coastline were observed by Imai et al. (2010) by
Dutykh and Dias 2009). The assumption of neglect- modeling delayed ruptures along the Nankai trough
ing temporal effects comes from the fact that tsunami where they caused the sub-faults to have a temporal
delay resulting in the worst case scenario. This
approach could be thought as the use of a variable
rupture velocity. Satake et al. (2013) performed a
1
Department of Geophysics, Faculty of Physical and Math- Multi time-window inversion to resolve the time
ematical Sciences, University of Chile, Santiago, Chile. E-mail:
mauricio@dgf.uchile.cl
history of the slip of the 2011 Tohoku-Oki earth-
2
National Seismological Center, Faculty of Physical and quake, with constant rupture velocity. They observed
Mathematical Sciences, University of Chile, Santiago, Chile.
 M. Fuentes et al. Pure Appl. Geophys.

that the best delayed slip model better explained the total rupture length and Vr is the rupture velocity. The
coastal tsunami heights than the instantaneous slip source duration and the rupture time are affected
model, which overestimated them. Fukutani et al. depending on the rupture mode: unilateral and bilat-
(2016) studied the uncertainties in tsunami estima- eral rupture propagation. Along-strike tsunami has a
tions due to kinematic rupture parameters. In faster velocity in the ocean-ward direction because
particular, they considered the rupture origin location the bathymetry is deeper in that direction. Rupture
and velocity along the strike direction. These kinds of modes are also important in tsunami directivity. It
studies are practical for probabilistic tsunami hazard was demonstrated by Bouchon (1980) that, in the
assessments. near field, there is a dynamic overshoot in the vertical
Regarding the analytical approach, Todorovska displacement field to up-dip propagation of the rup-
and Trifunac (2001) studied the problem of a hump ture for an inverse fault. If the lower layer has low
seafloor deformation with constant rupture velocity velocity value then the overshoot acquires an oscil-
and instantaneous source (no rise time). They found latory behavior. Geist (1998) simulated this effect
that wave amplification occurs when rupture velocity using the formula
is comparable with tsunami velocity. The attributed  
t
mechanism of amplification was wave focusing. On uz ðx; tÞ ¼ u0z 1  cosðx0 tÞe tR ;
the other hand, Dutykh and Dias (2007) modeled a
which is a ramp-like function including oscillations.
seafloor deformation with a finite rise time, but
Given that, the tsunami occurs in a much longer
simultaneously along the fault plane (instantaneous
period of time than tR .This should have a very small
rupture). Saito and Furumura (2009) included the
effect in the tsunami generation.
source duration to obtain a criterion, in terms of the
In this study, we will include rise time and rupture
fault dimensions and depth, for discriminating when a
time into analytical modeling in order to study the
dynamic generation should be preferred over a static
amplification on the tsunami generation and runup.
initial condition. Saito (2013) incorporated the rise
time to evaluate the influence of dynamic tsunami
generation over the ocean-bottom pressure evolution.
2. Mathematical Formulation
This is useful when using in combination with ocean-
bottom pressure gauges to estimate the water dis-
We consider the forced linear shallow water
placement. Nevertheless, all those studies were
equation in a sloping beach (Fig. 1):
applied in a constant depth ocean, with the 2 ? 1 D
linear potential theory. In this work, we trade-off one gtt  agðxgx Þx ¼ g0tt ; ð1Þ
spatial dimension for bathymetry complexity. We where gðx; tÞ is the water surface elevation, a ¼:
consider a sloping beach model, which is suitable for tanðbÞ is the slope of the beach, g is the gravity
earthquakes in subduction zones that triggers near-
field tsunamis.
Geist (1998) reviewed the effects of rise time,
rupture velocity, and dynamic overshooting. He used
the definition of t as a dimensionless number from
Hammack (1973). If this t  1, then the static dis-
placement is transferred immediately to the seafloor.
The other extreme t  1 is rare. He found that the
runup decreases if the rise time increases, and the
spatial variation of the rise time has almost no effects
on the tsunami generation. Rise time value is between
1 and 20 s for subduction earthquakes (Geist 1998).
The rupture time of the source is limited basically by Figure 1
Sketch of the sloping beach domain and variables of the 1 ? 1 D
two physical parameters: L and Vr , where L is the model
 Implications on 1 ? 1 D Tsunami Runup Modeling due to Time Features

acceleration, and g0tt ðx; tÞ is the forcing term applied

to the sea bottom.
Tuck and Hwang (1972) solved Eq. (1) by using
Hankel and Laplace transforms. The solution is
Z 1 Z 1 pffiffiffi
pffiffiffi
gðx; tÞ ¼ J0 ð2k xÞ J0 ð2k nÞg1tt ðn; tÞ (a)
1 1
pffiffiffiffiffi
sinð agktÞHðtÞdndk;

where
g1 ðx; tÞ ¼ g0 ðx; tÞ þ ½gðx; 0Þ þ tgt ðx; 0ÞHðtÞ;
HðtÞ is the Heaviside step function, J0 ðÞ is the zero-
(b)
order cylindrical Bessel function, and * denotes the
convolution product in time.
Figure 2
a Scheme of realistic setting of a subduction zone (Not to scale).
The rupture origin location is denoted by xR . b Source time
2.1. Source Time-Dependent Solution
function Tðx; tÞ for a given x
To model a non-instantaneous tsunami genera-
tion, we set null initial conditions, motion provided by the linear theory, gS ðtÞ ¼: gð0; tÞ,
gðx; 0Þ ¼ gt ðx; 0Þ ¼ 0; and g0 ðx; tÞ ¼ f0 ðxÞTðx; tÞ, can be written as
where f0 ðxÞ is the final shape of the seafloor 1
deformation and Tðx; tÞ 2 ½0; 1 is a temporal descrip- gS ðtÞ ¼ ½Mðf0 ; tV ÞðtÞ  Mðf0 ; tV Þðt  tR Þ: ð2Þ
tR
tion on how f0 ðxÞ is performed.T is directly related
with the temporal description of the seismic source From there, we can also obtain expressions for
function. We define quantifying individual effects of tR and Vr
  lim gS ðtÞ ¼ ot Mðf0 ; tV ÞðtÞ; ð3Þ
t  tV tR !0
Tðx; tÞ ¼ HðtR þ tV  tÞ þ Hðt  tR  tV Þ
tR
1
Hðt  tV Þ; lim gS ðtÞ ¼ ½Mðf0 ; 0ÞðtÞ  Mðf0 ; 0Þðt  tR Þ;
Vr !1 tR
ð4Þ
where tR is the rise time, tV ðxÞ ¼ jxx Rj
Vr is the rupture
time at location x for a bilateral rupture propagating where Eqs. (3) and (4) are for null rise time and
with a constant velocity Vr , starting from an origin xR infinite rupture velocity, respectively.
(Fig. 2). For simplicity, we consider tR constant. Note that
Computing Ttt in the sense of the distributions, we
lim gS ðtÞ ¼ lim gS ðtÞ
get tR ! 0 Vr ! 1
1 Vr ! 1 tR ! 0
Ttt ¼ ½dðt  tV Þ  dðt  tV  tR Þ: Z 1
tR 2 Hðagt2  4nÞ
¼ pffiffiffiffiffi ot pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi f0 ðnÞdn
ag 0 agt2  4n
Defining Z 1 1 2 

1o f0 4 agt y
Z ¼ t pffiffiffiffiffiffiffiffiffiffiffi dy ;
2 1
Hðag½t  sðnÞ2  4nÞ 2 ot 0 1y
Mðf; sÞðtÞ ¼ pffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi fðnÞdn;
ag ð5Þ
0 ag½t  sðnÞ2  4n
which is the same as the solution for the static case
and following the same procedure as in Fuentes derived by Fuentes (2017).It is also clear to observe
(2017) with each term, the approximated shoreline that
 M. Fuentes et al. Pure Appl. Geophys.

x  
lim gS ðtÞ ¼ lim gS ðtÞ ¼ 0: x  W cosðdÞ
Vr !0 tR !1 Uz ðxÞ ¼ uz  uz ; ð6aÞ
d d  W sinðdÞ
x  
x  W cosðdÞ
2.2. The Seabed Deformation Uh ðxÞ ¼ uh  uh : ð6bÞ
d d  W sinðdÞ
As mentioned, the final shape of the initial
condition is denoted by f0 ðxÞ. In tsunami modeling, Originally, the retained variable should be y, but
we renamed it as x to keep consistency, and it
the Okada’s equations are widely used to compute the
represents the variable along the half-space. uz and uh
static deformation due to a finite fault in 3D (Okada
are the dislocations defined as
1985). It provides a solution for the problem of static

deformation in an elastic half-space. Freund and U cosðdÞ  x sinðdÞ
uz ð
xÞ ¼ xÞ 
sinðdÞ arctanð
Barnett (1976) solved the 2D problem of surface p 1 þ x2
deformation and their solution can also handle non- 
uniform slip distributions. þ ð3  8mÞd sinðdÞ ;

ð7aÞ
2.2.1 Uniform Slip

U sinðdÞ þ x cosðdÞ
uh ð
xÞ ¼ xÞ 
cosðdÞ arctanð
We take a pure dip-slip (rake angle is 90°) fault of p 1 þ x2

width W, length L, dip angle d, slip U in a medium of
þ sinðdÞ þ d cosðdÞ  2ð1  2mÞd sinðdÞf2d tanðdÞ þ 1g :
Poisson ratio m.The fault is oriented parallel to the
coast, so the strike angle can be fixed, in our case, to ð7bÞ
180° due to the axis orientation chosen (Fig. 3a). In
(
x denotes a ratio between the horizontal distance and
the case of a pure dip-slip fault, the slip vector
depth of the fault endpoint) which are the same
reduces to ðU1 ; U2 ; U3 Þ ¼ ð0; U; 0Þ:
expressions obtained by Freund and Barnett (1976)
Both studies used different coordinate system, and
for uniform slip and corrected Madariaga (2003). It
then they will be reoriented to our reference system.
must be noted that constant terms in the previous
 LPlacing  the lower corner of the fault in equations were ignored by other authors. However, it
 2 ; 0; d and letting L tends to infinity, the
is not quite important because the final displacement
Okada’s solution for the seabed displacements can
Ui is the difference of dislocations ui at the endpoints
be reduced to the following expressions:

(a) (b)

Figure 3
a Geometry and spatial orientation of the fault plane [adapted from Okada (1985)]. b Profile of a dip-slip fault with slip distribution DðsÞ.
[adapted from Freud and Barnett (Freund and Barnett 1976)]
 Implications on 1 ? 1 D Tsunami Runup Modeling due to Time Features

  
of the fault (i ¼ z; h). Here, we will keep them for 1 x  s cosðdÞ
completeness. uz ðx; sÞ ¼ sinðdÞ arctan
p s sinðdÞ
In the worst case, the fault breaks up to the free surface 2  ð9aÞ
sxsin ðdÞ
of the half-space. It is well-known that a singularity is þ 2 ;
x  2sx cosðdÞ þ s2
produced, but it is still possible to retain a closed form.
  
This is obtained by letting d tends to W sinðdÞþ , 1 x  s cosðdÞ
 uh ðx; sÞ ¼ cosðdÞ arctan
x p s sinðdÞ
U  ð9bÞ
Uz ðxÞ ¼ sinðdÞ arctan ½s  x cosðdÞs sinðdÞ
p d þ 2 :
 x  2sx cosðdÞ þ s2
d½d cosðdÞ  x sinðdÞ
  C1 ðx  W cosðdÞÞ ;
x2 þ d 2 Then, for a general slip distribution DðsÞ and a
 x fault of width W buried at a distance a from the
U
U h ð xÞ ¼ cosðdÞ arctan origin, the final displacements are
p d
 Z aþW
d½x cosðdÞ þ d sinðdÞ Uz ðxÞ ¼ uz ðx; sÞD0 ðsÞds; ð10aÞ
  C2 ðx  W cosðdÞÞ ;
x2 þ d 2 a
Z aþW
where d ¼ W sinðdÞ,
Uh ðxÞ ¼ uh ðx; sÞD0 ðsÞds: ð10bÞ
p a
C1 ðzÞ ¼: sinðdÞsgn(zÞ þ cosðdÞ1f0g ðzÞ;
2 Freund and Barnett (1976) proposed, as example,
p the following normalized slip distribution:
C2 ðzÞ ¼: cosðdÞsgn(zÞ þ sinðdÞ1f0g ðzÞ;
2 ( 12
0 q3 ðq  rÞr; r\q
sgn(zÞ is the sign function and 1A ðzÞ is the indicator D ðrÞ ¼ 12 ; ð11Þ
ð1qÞ3
ðr  1Þðr  qÞ; r q
function.
Following Tanioka and Satake (1996), the com- where r ¼ saW and q 2 ð0; 1Þ.
plete vertical displacement of the ocean bottom d
In our coordinate system, a ¼ sinðdÞ  W and the
corresponds to the vertical component Uz ðxÞ plus final vertical displacement is
the horizontal advection contribution. Also, we must
f0 ðxÞ ¼ Uz ðx  xe Þ þ aUh ðx  xe Þ; ð12Þ
translate the solutions to our coordinate system.
Defining the trench axis location at x0 and with xe ¼ x0 þ a cosðdÞ.
xe ¼: x0  W cosðdÞ;, we finally obtain
f0 ðxÞ ¼ Uz ðx  xe Þ þ aUh ðx  xe Þ: ð8Þ
3. Numerical Tests
We can observe, when the term aUh is neglected,
that the maximum slip transformed into vertical 3.1. A Classical Mw ¼ 8:0 Earthquake
displacement occurs in the singularity (Ward 2011);
thus In this section, we will set d ¼ 20 being a typical
  value in the Chilean subduction zone (Hayes et al.
f0 ðx0 Þ d 2012). Once the dip angle is fixed, U and W are free
¼ sinðdÞ 1  :
U p to be chosen. Nonetheless, to keep realistic earth-
The optimal dip angle that maximizes the uplift is quake fault scales and to diminish the number of
parameters, we use the scaling laws from Blaser et al.
63:76 giving 58% of slip converted into vertical
(2010), assuming a constant rigidity of the medium
displacement.
¼ 30GPa. Then,
2.2.2 Non-uniform Slip logðWÞ ¼ 1:86 þ 0:46Mw ;

We call s to the local variable along the fault and logðUÞ ¼ 3:15 þ 0:47Mw ;
 M. Fuentes et al. Pure Appl. Geophys.

(a) (b)

(c) (d)

Figure 4
a Vertical displacement for a Mw ¼ 8:0 earthquake (Eqs. 6a–8) computed at the sea bottom. Inner plot depicts the normalized uniform slip
distribution used. b Vertical displacement for a Mw ¼ 8:0 earthquake (Eqs. 9a, 9b, 10a, 10b, 12) computed at the sea bottom. Inner plot
depicts the normalized skewed ðq ¼ 0:3Þ slip distribution used (Eq. 11). c Shoreline motion induced by a for different cases, with
Vr ¼ 2 km=s, xR ¼ 80 km (asymmetric bilateral rupture) and tR ¼ 15 s (Eqs. 2–5). d Shoreline motion induced by b for different cases, with
Vr ¼ 2 km=s, xR ¼ 80 km (asymmetric bilateral rupture) and tR ¼ 15s (Eqs. 2–5)

when comparing a combined effect of rupture

where Mw is the moment magnitude, W is in kilo-
velocity and rise time with a static initial condition.
meters, and D is in meters. This gives W
66 km
The case of non-uniform slip presents a 15% of
and D
4:1 m.
amplification. These calculations were made with
Figure 4 displays the shoreline motion due to a
values indicated in the caption of Fig. 4.
typical Mw ¼ 8:0 earthquake for a case with uniform
In Figs. 5 and 6, we display the complete
slip and another with non-uniform slip distribution
concentrated up-dip. One can see a clear difference in variation of the maximum runup relative to the static
case for models with rupture velocity and rise time
the deformation profile when assuming uniform or
for four modes of rupture, for two slip distributions:
non-uniform slip. For the values chosen in the
uniform and heterogeneous (Fig. 4). Figure 6 shows
modeling, the results show that in terms of the
that the amplification becomes very important for low
shoreline motion, it presents larger peak-to-peak
rupture velocities (of the order of the tsunami
amplitudes when considering the combined effect of
velocity). For very slow ruptures, the rise-time has
tR and Vr than in the static case. In the case of the
little affect on the maximum runup. For larger rupture
uniform slip fault, the amplification is 11% higher
 Implications on 1 ? 1 D Tsunami Runup Modeling due to Time Features

(a) (b)

Figure 5
Effect of the rise time in the runup height and time shift for an instantaneous rupture. a Variation of the normalized runup (with respect to the
static case) with variable rise time for homogeneous and heterogeneous slip distributions. The heterogeneous slip distribution was taken from
eq. (11) with q ¼ 0:3. b Same as a for the time shift relative to the peak runup time of the static case

velocities, when rise-time increases, the maximum 1

RtR ¼ ½Mðf0 ; tV Þðt1 Þ  Mðf0 ; tV Þðt1  tR Þ:
runup tends to be in the order of the value of the static tR
case. The complete behavior of these parameters will
Similarly, we call R0 the maximum runup in the
be examined in detail in the next section.
case of instantaneous rise of the seafloor deformation
and t0 the time where it is attained.
3.2. Parametric Analysis
R0 ¼ ot Mðf0 ; tV Þðt0 Þ:
There are many parameters involved in the whole
By virtue of the mean value theorem, there exist
problem of the tsunami runup; nonetheless, we focus
t 2 ðt1  tR ; t1 Þ such that
on those aspects related to the temporal evolution of
the source. Since the analytical solution is dependent RtR ¼ ot Mðf0 ; tV ÞðtÞ:
on the initial deformation profile, we keep fixed that
Since R0 is the maximum of the time series,
shape by using the same as one set in the previous
subsection. RtR  R0
which is true for any rise time. This result also shows
3.2.1 Rise Time
that the rise time introduces a time delay in the runup.
Figure (5) confirms the decreasing behavior of the
We call RtR the maximum runup induced by an initial
ratio RtR =R0 with increasing rise time (Fig. 5a) and
condition applied with a rise time tR , and t1 the time
also the trend that follows the time shift of the
where it is attained.
maximum runup relative to the static case (Fig. 5b).
This test was performed with infinite rupture velocity,
 M. Fuentes et al. Pure Appl. Geophys.

Figure 6
Amplification of the runup due to the rupture velocity considering uniform and non-uniform slip distributions. Each curve is associated with a
different rupture origin location xR . Both initial shapes were considered from subfigures (4a, b) with null rise time

which makes the origin location of the rupture xR and rise time. For instance, level curve ‘‘1’’ represents
irrelevant. the isocontour where the effects of the rupture
velocity and rise time are perfectly compensated.
3.2.2 Rupture Velocity Nevertheless, as it can be deduced from the previous
subsection, the worst-case scenario lies in the zone of
In this case, an analytical conclusion from the low rupture velocities (0.3–0.5 km/s) and null rise
equations is more complicated to obtain. Neverthe- time. Lower velocities were also tested (0.1–0.3 km/
less, we examine a wide range of rupture velocities s), but numerical treatments for the integration
where amplification is observed. Again, the results become quite complex, even though the runup is still
are compared with respect to the static case, and the amplified, as founded by Todorovska and Trifunac
rise time is set to zero in all these simulations. (2001). In general, for regular earthquakes, the
Figure 6 shows the behavior of the amplification rupture velocity varies around 2.0–2.5 km/s and the
of the runup, relative to the static case, as a function rise time about 1–20 s (Kanamori and Brodsky 2004;
of the rupture velocity. The variation with the rupture Geist, 1998). This means that according to our model,
origin location reveals that the maximum amplifica- the time dependency of the earthquake source is
tion occurs for a bilateral rupture that initiates at the responsible for 15–20% of the amplification.
middle of the fault. Also, the amplification is greater
for a rupture propagating downdip (updip origin)
instead the opposite. 4. Discussion and Conclusions

3.2.3 Combined Effect of Vr and tR Since continuous GPS and broadband seismome-
ter observations have increased, seismologists have
As a summary, Figs. (7, 8) present the whole
detected and observed several types of slow earth-
variation of the runup in terms of the rupture velocity
quakes or non-regular earthquakes such as slow slip
 Implications on 1 ? 1 D Tsunami Runup Modeling due to Time Features

(a) (b)

(c) (d)

Figure 7
Color map of the maximum runup relative to the static case from Fig. 4a for four different rupture origins. a xR ¼ 60 km (unilateral up-dip). b
xR ¼ 80 km (asymmetric bilateral). c xR ¼ 100 km (asymmetric bilateral). d xR ¼ 120 km (unilateral down-dip). Some level curves are
displayed in red for visual guidance

events (SSE), episodic and tremors, slip earthquakes giant earthquakes of Valdivia 1960 and Sumatra
(ETS) (e.g. Miller et al. 2002; ; Beroza and Ide 2011), 2004, both ruptures are very complex. The moment
silent earthquakes (Kanamori and Stewart 1979), and rate functions for both events have been subject of
tsunami earthquakes (Kanamori 1972) in certain study (Kanamori and Cipar 1974; Lay et al. 2005;
zones around the world. A key question is: Do giant Ammon et al. 2005). These tsunamigenic earthquakes
thrust tsunamigenic earthquakes produce slow rup- have shown slow rupture velocities; however, due to
ture (0.1–0.5 km/s) velocities? There is no strong instrumental limitations, it is not well understood nor
evidence of observations of such slow earthquake how slow they were, even the rupture area for both
rupture velocity on tsunamigenic events in subduc- events were not determined until months or years
tion zones. However, this is related to observational after the mainshock (Barrientos and Ward 1990; Stein
capabilities rather than no existence of such and Okal 2005).
phenomena. For the giant 2004 Sumatra earthquake some
Large tsunamigenic earthquakes often produce runup observations are still not well explained, not
large aftershocks immediately after the mainshock, just because of the limited resolution of the bathy-
then instruments are still reverberating for many metry or topography. During this event the Bay of
hours. This complicates our observations of such Bengal did not experience structural damage and
small wave amplitudes that carry information on slow intensities were documented at levels I and II, sug-
earthquake rupture velocity. Examples of this are the gesting the presence of a very slow rupture
 M. Fuentes et al. Pure Appl. Geophys.

(a) (b)

(c) (d)

Figure 8
Color map of the maximum runup relative to the static case from Fig. 4b for four different rupture origins. a xR ¼ 60 km (unilateral up-dip). b
xR ¼ 80 km (asymmetric bilateral). c xR ¼ 100 km (asymmetric bilateral). d xR ¼ 120 km (unilateral down-dip). Some level curves are
displayed in red for visual guidance

component considering the inundation reached there with this 1 ? 1 D solution we can still capture some
(Lay et al. 2005). In this work, we isolate the effect of overall features of the seismic source, and it has the
earthquake rupture velocity, rise time, and runup due advantage over previous studies that consider a
to a rupture propagating along the dip direction. We sloping beach bathymetry rather than a flat ocean and
computed the runup amplification due to a very slow the novelty of combining two temporal parameters at
moment rate in thrust earthquakes along the dip the same time.
direction. We are aware of other amplification runup We have demonstrated using a simple source
effects such as resonance in bays, shelf resonance, model that time evolution of slip can play an
and edge waves. These effects are not taken into important role in the tsunami modeling and its con-
account in our analytical formulation; instead, we sequent runup. It cannot be neglected for some cases,
explore the effects on the runup due to source especially when rupture velocities around 0:1 
parameters controlling the rupture kinematic. The 0:5 km=s can amplify the runup up to five times
tsunami amplitude is larger with a slower rupture compared to the static case. This suggests that the
velocity (0.1–0.5 km/s) than with a regular one system ocean-earth resonates with the tsunami wave
(2.0–2.5 km/s). periods (tsunami phase velocity rounds 0.2 km/s).
Due to the difficult equation treatment, we are still Mega-large earthquakes that present slow rupture
limited to obtain a solution including the source time velocities, which have generated enormous tsunamis
function in the 2 ? 1 D approximation. Nonetheless, in the near-field, might be conditioned by this slow
 Implications on 1 ? 1 D Tsunami Runup Modeling due to Time Features

rupture velocity component. This could be an Hammack, J. L. (1973). A note on tsunamis: their generation and
propagation in an ocean of uniform depth. Journal of Fluid
explanation for the low earthquake intensity and the
Mechanics, 60(04), 769–799.
abnormal runup heights in Bengal Bay during the Hayes, G. P., Wald, D. J., and Johnson, R. L. (2012). Slab 1.0: A
2004 Mw 9:3 Sumatra earthquake. Further studies three-dimensional model of global subduction zone geometries.
taking into account the combined effects of ruptures Journal of Geophysical Research: Solid Earth, 117(B1), B01302.
https://doi.org/10.1029/2011JB008524.
in the strike and dip directions are necessary to Ide, S., Beroza, G. C., Shelly, D. R., & Uchide, T. (2007). A scaling
quantify accurately the amplification factors. law for slow earthquakes. Nature, 447(7140), 76–79.
Imai, K., Satake, K., & Furumura, T. (2010). Ampli_cation of
tsunami heights by delayed rupture of great earthquakes along
the Nankai trough. Earth, Planets and Space, 62(4), 427–432.
Acknowledgements Kajiura, K. (1970). Tsunami source, energy and the directivity of
wave radiation. Bulletin of the Earthquake Research Institute, 48,
835–869.
This work was entirely funded by the Programa de
Kajiura, K. (1981). Tsunami energy in relation to parameters of the
Riesgo Sı́smico. earthquake fault model. Bulletin of the Earthquake Research
Institute, 56, 415–440.
Kanamori, H. (1972). Mechanism of tsunami earthquakes. Physics
REFERENCES of the Earth and Planetary Interiors, 6(5), 346–359.
Kanamori, H., & Brodsky, E. E. (2004). The physics of earth-
Ammon, C. J., Ji, C., Thio, H. K., Robinson, D., Ni, S., Hjor- quakes. Reports on Progress in Physics, 67(8), 1429.
leifsdottir, V., et al. (2005). Rupture process of the 2004 Kanamori, H., & Cipar, J. J. (1974). Focal process of the great
Sumatra–Andaman earthquake. Science, 308(5725), 1133–1139. Chilean earthquake May 22, 1960. Physics of the Earth and
Barrientos, S. E., & Ward, S. N. (1990). The 1960 Chile earth- Planetary Interiors, 9(2), 128–136.
quake: inversion for slip distribution from surface deformation. Kanamori, H., & Stewart, G. S. (1979). A slow earthquake. Physics
Geophysical Journal International, 103(3), 589–598. of the Earth and Planetary Interiors, 18(3), 167–175.
Beroza, G. C., & Ide, S. (2011). Slow earthquakes and nonvolcanic Kânoğlu, U. (2004). Nonlinear evolution and runup-rundown of
tremor. Annual Review of Earth and Planetary Sciences, 39, long waves over a sloping beach. Journal of Fluid Mechanics,
271–296. 513, 363–372.
Blaser, L., Krüger, F., Ohrnberger, M., & Scherbaum, F. (2010). Lay, T., Kanamori, H., Ammon, C., Nettles, M., Ward, S., Aster,
Scaling relations of earthquake source parameter estimates with R., et al. (2005). The great Sumatra–Andaman earthquake of 26
special focus on subduction environment. Bulletin of the Seis- December 2004. Science, 308(5725), 1127–1133.
mological Society of America, 100(6), 2914–2926. Madariaga, R. (2003). Radiation from a finite reverse fault in a half
Bouchon, M. (1980). The motion of the ground during an earth- space. Pure and Applied Geophysics, 160, 555–577.
quake: 2. The case of a dip slip fault. Journal of Geophysical Madsen, P. A., & Schaffer, H. A. (2010). Analytical solutions for
Research: Solid Earth, 85(B1), 367–375. tsunami runup on a plane beach: single waves, N-waves and
Carrier, G. F., & Greenspan, H. P. (1958). Water waves of _nite transient waves. Journal of Fluid Mechanics, 645, 27–57.
amplitude on a sloping beach. Journal of Fluid Mechanics, 4(01), Miller, M. M., Melbourne, T., Johnson, D. J., & Sumner, W. Q.
97–109. (2002). Periodic slow earthquakes from the Cascadia subduction
Dutykh, D., and Dias, F., (2007). Water waves generated by a zone. Science, 295(5564), 2423.
moving bottom. In Tsunami and Nonlinear waves. In Tsunami Okada, Y. (1985). Surface deformation due to shear and tensile
and Nonlinear waves, 65-95. Springer Berlin Heidelberg. faults in a half-space. Bulletin of the Seismological Society of
Dutykh, D., & Dias, F. (2009). Tsunami generation by dynamic America, 75, 1135–1154.
displacement of sea bed due to dip-slip faulting. Mathematics Saito, T. (2013). Dynamic tsunami generation due to sea-bottom
and Computers in Simulation, 80(4), 837–848. deformation: Analytical representation based on linear potential
Freund, L. B., & Barnett, D. M. (1976). A two-dimensional anal- theory. Earth, Planets and Space, 65(12), 1411–1423.
ysis of surface deformation due to dip-slip faulting. Bulletin of Saito, T., & Furumura, T. (2009). Three-dimensional tsunami
the Seismological Society of America, 66(3), 667–675. generation simulation due to sea-bottom deformation and its
Fuentes, M. (2017). Simple estimation of linear 1 ? 1 D long wave interpretation based on the linear theory. Geophysical Journal
run-up. Geophysical Journal International, 209(2), 597–605. International, 178(2), 877–888.
Fuentes, M., Ruiz, J., & Cisternas, A. (2013). A theoretical model Satake, K., Fujii, Y., Harada, T., & Namegaya, Y. (2013). Time
of tsunami runup in Chile based on a simple bathymetry. Geo- and space distribution of coseismic slip of the 2011 Tohoku
physical Journal International, 196(2), 986–995. earthquake as inferred from tsunami waveform data. Bulletin of
Fukutani, Y., Anawat, S., & Imamura, F. (2016). Uncertainty in the Seismological Society of America, 103(2B), 1473–1492.
tsunami wave heights and arrival times caused by the rupture Stein, S., & Okal, E. (2005). Speed and size of the Sumatra
velocity in the strike direction of large earthquakes. Natural earthquake. Nature, 434(7033), 581–582.
Hazards, 80(3), 1749–1782. Synolakis, C. E. (1987). The runup of solitary waves. Journal of
Geist, E. L. (1998). Local tsunamis and earthquake source Fluid Mechanics, 185, 523–545.
parameters. Advances in Geo-physics, 39, 117–209.
 M. Fuentes et al. Pure Appl. Geophys.

Tanioka, Y., & Satake, K. (1996). Tsunami generation by hori- Tuck, E. O., & Hwang, L.-S. (1972). Long wave generation on a
zontal displacement of ocean bottom. Geophysical Research sloping beach. Journal of Fluid Mechanics, 51, 449–461.
Letters, 23(8), 861–864. Ward, S. (2011). Tsunami. In: Gupta H.K. (eds) Encyclopedia of
Todorovska, M. I., & Trifunac, M. D. (2001). Generation of tsu- Solid Earth Geophysics. Encyclopedia of Earth Sciences Series.
namis by a slowly spreading uplift of the sea oor. Soil Dynamics Springer, Dordrecht.
and Earthquake Engineering, 21(2), 151–167.

(Received June 22, 2017, revised January 20, 2018, accepted February 7, 2018)