Medina2021 Article TsunamiModelingInTheSouthAmeri

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Pure Appl. Geophys.

Ó 2021 The Author(s), under exclusive licence to Springer Nature Switzerland AG


https://doi.org/10.1007/s00024-021-02808-w Pure and Applied Geophysics

Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling
and Historical Seismicity
MIGUEL MEDINA,1 SEBASTIÁN RIQUELME,2 MAURICIO FUENTES,1 and JAIME CAMPOS1

Abstract—Throughout its history, South America has experi- (M * 9.0) (Carvajal et al., 2017a; Udı́as et al., 2012),
enced megathrust earthquakes which have produced large
tsunamis, devastating coastal cities in the near and far field.
1868 in southern Peru (M * 8.8) (Comte & Pardo,
Studying these phenomena is important for tsunami hazard miti- 1991), 1877 in northern Chile (M * 8.8) (Comte &
gation in this region. We propose 10 earthquake scenarios along the Pardo, 1991), 1906 in Ecuador–Colombia (Kanamori
South American subduction zone on the basis of the seismic history
& McNally, 1982), and 1960 in southern Chile
of each region, seismic-geodetic coupling, general scaling rela-
tionships, among others. Tsunami run-up (coastal amplification) is (M * 9.5) (Kanamori & Cipar, 1974), among others
then estimated using 200 nonuniform stochastic sources for each (see Fig. 1).
scenario, totaling 2000 simulations. Our results show great vari- Learning from these past events is essential to
ability in run-up distribution along the Nazca-South America
subduction zone, with some of the most affected areas being Val- improve our comprehension of tsunamis to identify
paraiso in Chile, with a most likely scenario of 20 m of run-up and hazards. Then, we can reduce the disruptive impacts
a maximum scenario of 33 m; and Lima in Peru, with 25 and 40 m of a natural disaster on communities along the coast
for the most likely and maximum scenarios, respectively. Similar
results are seen in Iquique and Huasco in Chile. We have also
of South America.
identified 17 coastal locations with a higher vulnerability due to For tsunami hazard assessment, there are different
local amplification of tsunami run-up and two instances in which a methodologies that can be classified based on the
regional amplification can occur due to tsunami directivity and
distance to the tsunami source: far- and near-field. In
coastal barriers. We conclude that tsunami hazard remains high
along the coast of South America, even in areas where great the far-field case, a simple model as a function of the
earthquakes have recently occurred. earthquake magnitude is often enough for tsunami
hazard assessment. Slip distribution and source depth
Keywords: Tsunami, South America, seismology.
can even be seen as second-order factors, while other
phenomena such as directivity play a more important
role (Okal, 1988).
On the contrary, for near-field tsunamis, slip dis-
1. Introduction
tribution, source geometry, and focal mechanism play
a key role and, are of the utmost importance when
Tsunamis in South America can be traced back
modeling tsunamis on the coast (Geist, 2002). As
centuries using both historical records and paleo-
previously documented, this is why uniform slip
tsunamis (Cisternas et al., 2005; Goff et al., 2020;
sources often underestimate the maximum vertical
Spiske et al., 2013). These events have had devas-
inundation (Fuentes et al., 2019; Geist & Dmowska,
tating consequences, among which the most
1999; Ruiz et al., 2015) which is known as run-up.
important historical tsunamigenic earthquakes in
Strictly speaking, this parameter can only be modeled
South America occurred in 1730 in central Chile
by numerical methods with high-resolution
bathymetries. This data is not always available, and
approximations based on analytical or empirical
1
Seismic Risk Program, University of Chile, Santiago, models are adopted. Hereafter, we will treat the run-
Chile. E-mail: miguel.medina@dgf.uchile.cl; mauricio@dgf.
uchile.cl; jaime@dgf.uchile.cl
up as an approximation from the peak tsunami
2
National Seismological Center, University of Chile, nearshore amplitude (PNTA). Typical water depths
Santiago, Chile. E-mail: sebastian@csn.uchile.cl
M. Medina et al. Pure Appl. Geophys.

(where this value is computed) span between 50 and Figure 1 c


a Seismic history in South America, estimated and measured
200 m. In this study, we will take the nearest grid rupture lengths along the coast for interplate earthquakes. Stars
point to the grid coastline (the change of sign in the denote non-interplate earthquakes (intraplate and tsunami earth-
bathymetry data). quakes), and grey bars denote bathymetric relief such as ridges and
fracture zones. b South American subduction zone. Earthquake
There is a lack of regional-scale studies with a
geometry and magnitude are plotted for each modeled scenario
standard methodology, which would allow for com- along the subduction zone
parison along the South American subduction zone.
Tsunami hazard studies in South America are often
done on a small scale, because a detailed bathymetry is
needed to model them properly. This has been done due Gutenberg–Richter law (G–R), and scaling laws for
the efforts of local governments in specific coastal certain scenarios. For each of these 10 scenarios, we
areas or to study historical tsunamis (Carvajal et al., only define a moment magnitude and a fault geom-
2017a; Okal et al., 2006). Recent studies have been etry; meaning that slip distribution is initially
focused on stochastic or hybrid approaches to assess unknown and is then determined by the use of
tsunami hazard (Becerra et al., 2020; Drápela et al., stochastic sources.
2021; Escobar et al., 2020; González et al., 2020; Mas The methodology proposed by Ruiz et al. (2015)
et al., 2014), moving away from the use of uniform is used for source magnitude and geometry estimation
distributions obtained from historical sources. These as well as tsunami modeling in each region, in which
have been known to underestimate run-up values up a magnitude is estimated by the particular seismic
to * 6 times for the maximum value (Geist, 2002). history of a region and stochastic sources are then
In this work, we study run-up along the South used to assess tsunami run-up with numerical tsunami
American coast with a regional approach to estimate modeling. A nonplanar source geometry is used for
thresholds for maximum amplitudes to be expected these sources. The Fuentes et al. (2019) approach for
along this region. We define 10 regional tsunami estimating an earthquake magnitude and size is also
scenarios in South America, each one consisting of employed, which expands the magnitude estimation
200 simulations for nonuniform stochastic sources. of Ruiz et al. (2015) by using other factors such as
Numerical tsunami modeling is conducted to obtain seismic coupling obtained from the seismic sequence
run-up distributions along this convergent tectonic of the region based on the work of Scholz and
boundary. Campos (1995, 2012).
For South America, we highlight eight zones of To define the extension and magnitude of each of
interest, each one with a defined seismic scenario, from the 10 earthquake scenarios, we estimate earthquake
south to north: Central Chile, North-Central Chile, return periods, seismic cycle duration, coupled area
Northern Chile, South Peru, Central Peru, Northern size, plate convergence rate, and shear modulus.
Peru, Ecuador–Colombia, and Colombia. As opposed These features are used to estimate the seismic cou-
to the other regions, in Northern Chile we explore three pling of each region following the work of Scholz
possible earthquake scenarios: a massive singular and Campos (1995, 2012), and finally an average
megathrust rupture in the whole region and two smaller expected magnitude is obtained for each scenario.
ruptures north and south, which are denoted as For each region, the estimation of return period is
Northern Chile, Northern Chile Up, and Northern based on the seismic analysis of the Gutenberg–
Chile Down, respectively. Thus, we define 10 seismic Richter law (Gutenberg & Richter, 1944) and the
scenarios in total which can be seen in Fig. 1. interseismic coupling degree. We use catalogs from
the US Geological Survey with a completeness
threshold of Mw 4.0 and a depth up to 50 km for each
2. Methods and Data of the studied regions. Large historical interplate
events (Mw [ 7.5) are also included and can be seen
The proposed scenarios are based in each region’s in Fig. 1. Return periods are estimated for large
seismic historical records, seismic coupling, the events to validate the proposed earthquake
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling
M. Medina et al. Pure Appl. Geophys.

Table 1
Estimated seismic cycles for different scenarios

Scenario ID Seismic cycle beginning Cumulative moment Released moment Available moment Associated magnitude
year (N m) (N m) (N m) (Mw)

Central Chile 1730 5.25e22 4.07e22 1.18e22 8.6


Northern-Central 1922 2.36e22 8.02e21 1.56e22 8.7
Chile
Northern Chile 1877 1.69e22 1.05e21 1.59e22 8.7
Down
Northern Chile Up 1868 9.11e21 5.11e21 4.00e21 8.3
Northern Chile 1877 3.86e22 4.94e21 3.36e22 9.0
Southern Peru 1868 1.25e22 5.61e21 6.87e21 8.5
Central Peru 1746 3.60e22 3.75e21 3.22e22 8.9
Ecuador– 1906 1.11e22 1.91e21 9.17e21 8.6
Colombia
The cumulative moment corresponds to the seismic moment accumulated from the seismic cycle beginning year up to 2020. The released
moment column corresponds to the seismic moment released since the cycle beginning by subduction earthquakes, and the available moment
is the difference between cumulative and released seismic moment with its associated magnitude moment in the next column. Northern Peru
and Colombia scenarios are not calculated due to lack of seismic data

magnitude, in the context of each seismic cycle and With these values, seismic coupling is computed
for seismic coupling calculation. using the following formula from Scholz and Campos
Seismic cycle periods are chosen from the last (2012):
largest earthquake up to year 2020 as detailed in 
Xs ¼ Moi = lvo Ac T i ; ð1Þ
Table 1. Return periods are obtained directly from the
G–R law. Different seismic cycle magnitudes are where Xs is the seismic coupling coefficient, Moi is the
tested and validated with these return periods for this moment released by the ith event with recurrence
newly defined seismic cycle. The use of the G–R law time T i , l is the shear modulus, vo is the plate con-
in this work is different from its usual role such as in vergence rate, and Ac is the coupled area. Values
a probabilistic tsunami hazard assessment as seen in obtained for seismic coupling derived from this
Geist and Parsons (2006) and De Risi and Goda simple methodology are validated with the ones
(2016). obtained by other authors using GPS networks in the
For the coupled area, we define a simple rectan- region (Béjar-Pizarro et al., 2013; Chlieh et al., 2011;
gular geometry composed of length and width. Métois et al., 2012, 2013, 2016; Mothes et al., 2018;
Length is estimated in each gap with several kilo- Vigny et al., 2009; Villegas-Lanza et al., 2016a).
meters towards north and south of the defined gaps, in Table 1 shows the estimated seismic deficit or
order to assure that the entire coupled region is accumulated seismic moment, which is computed by
contained. Width is estimated using the mean sub- multiplying the seismic flux defined as vo Ac by the
duction angle for each region and a coupled depth of elapsed time since the year the cycle began up to the
50 km, based on the proposed coupled depth from year 2020. The released energy by recent or historical
other previous works (Lay et al., 2012; Pacheco et al., events since the cycle beginning is accounted for the
1993; Tichelaar & Ruff, 1993). available moment and is calculated as the residual of
Plate convergence rates are obtained from Argus this energy and the accumulated seismic moment for
and Gordon (1991), Kendrick et al. (1999), and every region.
Trenkamp et al. (2002) for each particular region. We Nonplanar fault geometry is obtained from a
use a mean value of 32 GPa for the shear modulus forward modeling following Slab 2.0 (Hayes et al.,
based on the preliminary reference Earth model 2018) (see Fig. 2), using a subfault size close to
(PREM) model (Dziewonski and Anderson, 1981). 2 9 2 km to assure that each subfault can be
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 2
Fault geometry for the North-Central Chile scenario. a Tsunami modeling domain and surface deformation computing domain in rectangle.
b Fault geometry in dotted line, and profiles A–D are solid lines. Profiles A–D in depth and fault geometry agree with Slab 2.0

modeled as a point source. In each of the 10 sce- [0, 1]. We will take H ¼ 1, which makes the
narios, 200 sources or simulations are used for wavenumber spectrum decay with a power law k2 ,
tsunami modeling. This number is suggested by De usually referred to as the k2 model. The Von Karman
Risi and Goda (2016), who verified the stabilization filter allows for smoothly passing from low (flat
in the convergence of the wave height by simulating spectrum) to high (k2 decay) wavenumber domains.
500 scenarios in Japan. Then, samples are drawn by adding a random phase
In order to generate realistic earthquake sources, spectrum and emulating a slip distribution (Graves &
there are usually two equivalent approaches: assum- Pitarka, 2010). This approach has already been used
ing a probability distribution of the slip in the fault in Chile (Fuentes et al., 2019; Ruiz et al., 2015).
and proposing a correlation matrix, or imposing a Each scenario consists of two domains: the tsu-
specific spatial power spectrum of the slip. The nami propagation and the surface deformation
seminal works on both sides (Herrero & Bernard, domains. The surface deformation region is always
1994; Mai & Beroza, 2002) provide a framework on contained in the propagation domain; we call this the
this subject. In both cases, correlation lengths or internal region.
corner frequencies are needed (they are directly In the internal region, the source surface defor-
related by the scaling properties of the Fourier mation is computed using Okada’s point-source
transform). Usually, those correlation lengths are equations for sufficiently small fault discretization
modeled as a fraction of their dominant scales. (Okada, 1985). This region is smaller than the tsu-
Melgar and Hayes (2019) proposed new scaling nami modeling region, given that the surface
relations of correlation lengths and effective size of deformation decays quickly to zero in areas distant to
the fault based on a bigger finite fault model catalog, the source. Using this approach, computing times are
spanning more magnitudes and earthquake variabili- reduced. Geist (2002) pointed out that the grid size in
ties than Mai and Beroza (2002). Adopting this the fault should be less than or equal to the source
scaling relation for the correlation lengths, we impose depth, in order to correctly simulate waves at that
a power spectrum following a Von Karman filter, scale.
which depends on the correlation lengths and a con- This static surface deformation is then used as the
stant H, namely the Hurst exponent. This value lies in initial condition for numerical tsunami modeling.
M. Medina et al. Pure Appl. Geophys.

Static deformation sources are often used for As a factor that might be related to these
numerical tsunami modeling because a passive anomalous peak run-ups, we study the beach slope.
propagation is enough to represent regular subduction Using GEBCO 30-arc-second bathymetry and a
events, omitting slow events or tsunami earthquakes. piecewise linear approximation of bathymetry pro-
Tsunamis are a long-wave phenomenon, meaning files perpendicular to the coast, we obtain the angle of
that the total rupture time for a typical subduction the nearest segment to the shore which we denote as
earthquake is much smaller than the period of waves the beach slope b.
propagated in water. Finally, in the following sections, we delineate
Numerical tsunami modeling is computed using important background information used to determine
NEOWAVE (Non-hydrostatic Evolution of Ocean the size and magnitude for each of the regional sce-
WAVE) (Yamazaki et al., 2009). NEOWAVE narios proposed in this work, such as historical
numerically solves the nonlinear shallow water seismicity, notorious bathymetric features of the
equations with a non-hydrostatic pressure field in a region, among others. Scenarios are described from
semi-implicit finite difference scheme in a staggered south to north.
grid. This solver employs the upwind scheme to
account for advective terms and the non-hydrostatic
pressure term to account for weakly dispersive 3. Chile
waves. GEBCO 30-arc-second (* 925 m) bathy-
metry (Becker et al., 2009) is used with a vertical We define three regions of interest in Chile:
wall condition on the coast (that is to say, full Central Chile at around 32.5° S–37° S, Central-North
reflection at the coast), meaning that no flooding is Chile at around 28° S–32.5° S, and Northern Chile at
modeled. For a regional study such as this one, fine 20° S–28° S.
details of the ocean bottom are not necessary, and the
use of this bathymetry greatly reduces the modeling
times. Since only large earthquakes are modeled here 4. Central Chile
(Mw [ 8.0), this bathymetry grid size is safely much
lower than the slip correlation lengths used in this The Central Chile region is included from the
study (fault patch sizes). south of the Juan Fernandez Ridge (JFR, 32.5° S)
Once tsunami modeling is done, run-up distribu- down to the north of the Mocha Fracture Zone
tions can be constructed. For each seismic scenario, (MOFZ) at 37° S. The most recent event inside this
we construct run-up conditional probability of region is the 2010 Mw 8.8 earthquake (Delouis et al.,
exceedance based on the values obtained along the 2010; Vigny et al., 2011) in the located southern part,
coast for each set of 200 simulations. This is esti- and the largest one occurred in 1730 with an esti-
mated as the probability for run-up to be greater or mated magnitude of 9.1–9.3 (Carvajal et al., 2017a).
equal than a given value at a step of 0.5 m. The Although there have been multiple earthquakes in
probability is computed individually for each grid this region, the seismic deficit in the northern part has
point in the tsunami modeling coast at a 30-arc-sec- not been completed yet. This agrees with the seismic
ond resolution. We display these 2D probabilities in a coupling measurements, showing that this region is
heat color map for each scenario in Fig. 3. highly coupled (Métois et al., 2016). The 1906 Mw
For anomalous peaks in run-up distribution, far 8.2 earthquake (Comte et al., 1986) generated a very
from the source, we define an amplification factor moderate tsunami (Carvajal et al., 2017b), and the
f. This factor is calculated as the peak run-up divided 1985 Mw 7.8–8.0 event took place in the lower part
by mean run-up, in the immediate neighborhood in a of the subduction interface (Barrientos, 1988; Bravo
0.25° radius ball centered at this anomalous run-up. et al., 2019); this means that most of the seismic
This value estimates how much run-up has increased moment might be accumulated in the up-dip portion
in certain locations. of this region, enhancing the tsunami potential.
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 3
Obtained run-up exceedance probability distribution by latitude for the following scenarios: a Central Chile (Mw 8.6), b North-Central Chile
(Mw 8.7), c Northern Chile (Mw 9.0), d Northern Chile Down (Mw 8.5), e Northern Chile Up (Mw 8.3), f Southern Peru (Mw 8.5), g Central
Peru (Mw 8.9), h Northern Peru (Mw 8.8), i Ecuador–Colombia (Mw 8.6), and j Colombia (Mw 8.7). For i and j, distribution by longitude is
appended, and a dotted line is plotted in this separation. Earthquake source areas are highlighted in grey, and locations of interest are marked
M. Medina et al. Pure Appl. Geophys.

5. North-Central Chile to be nearly or fully coupled, as shown by many


authors (Béjar-Pizarro et al., 2013; Métois et al.,
North-Central Chile is the region starting at the 2013).
north of the Juan Fernandez Ridge up to the Taltal
Ridge (TR, 25° S). This region has a moderate to high
coupling (Métois et al., 2016). In the southern section 7. Peru
of this region, the most recent event was the 2015
Mw 8.3 Illapel earthquake (Fuentes et al., 2016; Ruiz For Peru there are three regions: Southern around
et al., 2016). Historically, the northern section has 15° S–18° S, Central at 10° S–15° S, and Northern
had at least two large earthquakes (M [ 8), once in Peru at 2.5° S–10° S.
1819 and once in 1922, both with an estimated
magnitude around Mw 8.5–8.6 (Carvajal et al.,
2017b). The latter had a significant regional tsunami 8. Southern Peru
registered (Lomnitz, 2004) as well as a transoceanic
tsunami recorded in Japan (Abe, 1979). If the rupture Southern Peru is defined as the area between the
area shared between the 1819 and 1922 earthquakes Nazca (NFZ, 15° S) and Iquique Ridges, with the
is extended with the rupture area of the 1918 Mw 7.5 biggest earthquake being the 1868 Mw 8.8 Arica
(Lomnitz, 2004) event, a 400-km-long rupture zone is earthquake (Comte & Pardo, 1991), similar to the
obtained between 30° S and south of the 1995 Mw 1877 earthquake just south of this megathrust event.
8.0 earthquake (Delouis et al., 1997), which is in Since this earthquake, the seismic gap has been par-
agreement with a Mw 8.5–8.8 event according to tially filled with the 2001 Mw 8.4 Pisco earthquake
scaling relationships (Blaser et al., 2010; Strasser (Bilek & Ruff, 2002), which had a considerable tsu-
et al., 2010). nami with run-ups of almost 10 m. This region
remains with a high coupling in the north (Villegas-
Lanza et al., 2016a).
6. Northern Chile

Northern Chile is one of the most studied seismic 9. Central Peru


gaps in the world (McCann et al., 1979), expanding
from the northern part of the Iquique Ridge (IR, 19° Located between the Mendaña Fracture Zone
S) down south to the Taltal Ridge. This region gained (MFZ, 10° S) and the Nazca Ridge, Central Peru is
notoriety for the absence of significant seismicity in without a doubt one of the most seismically active
nearly 140 years, from the years 1877–2014. The regions in the past century, having at least three
1877 Mw 8.8 earthquake (Comte & Pardo, 1991; earthquakes with a magnitude above 8.0: 1940 Mw
Lomnitz, 2004) is one of the biggest historical 8.0–8.2 (Beck & Nishenko, 1990; Kanamori, 1977),
earthquakes documented in Chile and had an 1974 Mw 8.1 (Beck & Nishenko, 1990), and 2007
immense tsunami (21 m in Mejillones) that caused Mw 8.0 (Sladen et al., 2010). And still, the seismic
devastation in nearby cities (Abe, 1979; Lomnitz, moment for these three events only partially fills the
2004). Though the more recent 2014 Mw 8.1 earth- gap since the 1746 Mw 8.6–9.0 (Dorbath et al., 1990)
quake was big enough to cause damage (Becerra earthquake. If the seismic deficit is calculated from
et al., 2016), the consequential tsunami was moderate this event up until now, there is enough seismic
and mostly instrumental (Catalán et al., 2015), and energy for a Mw 8.9 event (Villegas-Lanza et al.,
thus most of the damage associated with this earth- 2016a).
quake was related to shaking. This earthquake failed
to fill the enormous seismic gap in the region (Hayes
et al., 2014), and seismic hazard remains high in this
region. Most of the Northern Chile subduction seems
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

10. Northern Peru suggests a seismic super-cycle happening in this


region, similar to the Sumatra earthquake cycle (Sieh
From the northernmost coastal point of Peru et al., 2008), meaning that there may be more seismic
roughly south of the Carnegie Ridge (CAR, 2.5° S) moment in the region to be released, more than that
down to the Mendaña Fracture Zone, Northern Peru accumulated since 1906. The northern and central
is a region lacking significant seismicity and is segments are shown to be highly coupled (Mothes
mostly associated with aseismic movements (Noc- et al., 2018). The seismic deficit could be high if there
quet et al., 2014; Villegas-Lanza et al., 2016b) and is indeed a super-cycle happening in this region.
slow or tsunami earthquakes (Pelayo & Wiens, 1990).
The only historical earthquake occurred in 1618
(Bourgeois et al., 1999; Dorbath et al., 1990), and its 13. Colombia
mechanism is not well defined, meaning that it could
have been an intraplate earthquake. Seismic coupling Lastly, the Colombia region is located north of the
is low for most of the region, but is poorly con- 1906 rupture area and up to the border with Panamá
strained north of the Medaña Fracture Zone near the at * 7° N. In this area there is a lack of significant
trench (Villegas-Lanza et al., 2016a). seismic history, which could be due to the lack of
considerable population in this rainy area or due to
the change from a subduction to a transform regimen
11. Ecuador and Colombia in the northern part of this region (see Fig. 1). This
last option does not rule out the possibility of an
Ecuador and Colombia have been paired given earthquake extending up until this regimen change,
that most of their seismic history is shared, as with similar to the Mw 9.5 1960 Valdivia (Moreno et al.,
the 1906 Mw 8.8 earthquake rupturing in both 2009) earthquake extending from the Nazca plate
countries. Two regions for modeling are defined: down to the Antarctic plate, where these two plates
Ecuador–Colombia (1° S–4° N) and Colombia (4° N– have a transform boundary. Most of the seismic
7° N). network is inland (Vargas et al., 2018), and we could
not find a seismic coupling model for this area.

12. Ecuador–Colombia
14. Results
The Ecuador–Colombia region is defined as north
of the Carnegie Ridge (CAR, 2.5° S) up until the end Estimated magnitudes for the proposed scenarios
of the rupture for the 1906 earthquake at * 4° S. are detailed as follows: (1) Chile Central: a moment
This region has one of the most studied and under- magnitude 8.6 earthquake is estimated for this region
stood seismic sequences, starting with the 1906 Mw from the northern part of the 2010 earthquake rupture
8.8 earthquake rupturing the entire region. In this to the southern portion of the 2015 earthquake rup-
region, there are other three rupture models (Kana- ture. (2) North-Central Chile: we propose a
mori & McNally, 1982): the southern segment magnitude 8.7 scenario for this area, based on seismic
ruptured in 1942 and later in 2016 (Ye et al., 2016) deficit estimation. (3–5) Northern Chile: here we
with both events having similar magnitudes and explore two possibilities, a big rupture in the same
rupture areas. The central segment ruptured in 1958, area of the 1877 earthquake with a magnitude 9.0
and the northern segment ruptured in 1979. However, (Northern Chile), and two smaller ruptures south and
the 2016 Mw 7.8 earthquake ruptured in the southern north of the 2014 Iquique earthquake, with magni-
segment similar to the 1942 event, so it is possible tudes Mw 8.3 (Northern Chile Down) and 8.5
that the seismic deficit was exceeded (Nocquet et al., (Northern Chile Up), respectively, the last two com-
2017) in this region. The 1958 and 1979 events also bined being smaller than the accumulated seismic
seem to exceed the expected magnitude. This moment in the region. (6) Southern Peru: the
M. Medina et al. Pure Appl. Geophys.

proposed earthquake scenario is a Mw 8.5 earth- ups are mostly below 5 m, and results are similar
quake, north of the 2001 event and partially above the along different simulations.
Nazca Ridge. (7) Central Peru: a Mw 8.9 magnitude In particular, some of the most significant results
event is modeled, which is the second largest mod- are as follows: Valparaı́so in Chile (33° S) with most
eled earthquake only behind the Northern Chile Mw likely run-ups exceeding 10 to 15 m and maximum
9.0 scenario. (8) Northern Peru: an event is modeled well above 30 m, similar to what we see in Huasco
in the worst GPS coverture area near the trench, and a (28.5° S). In Central Peru we see the highest run-up
magnitude Mw 8.8 is obtained using only scaling values in the country with some surpassing 40 m, and
relationships for the size of the selected area. (9) a significant portion of the coast (10° S–12° S) with
Ecuador–Colombia: for the chosen earthquake, we most probable scenarios in the 10–15-m range.
combine the central and northern segment, where Northern Peru has a very narrow distribution between
there is enough space for a Mw 8.6 earthquake in an Chiclayo (6.5° S) and Trujillo (8° S), with maximum
area shown to be highly coupled. (10) Colombia: run-up values in excess of 20 m. Ecuador and
without any data to rule out this model, a Mw 8.7 Colombia show lower run-up values mostly below
earthquake is modeled in the entirety of this region the 10-m threshold, but these distributions seem to
obtained using scaling relationships. Rupture sizes expand over broader areas.
and magnitudes can be seen in Fig. 1.
For each scenario, there are 200 simulations of
different stochastic distributions, each producing a 15. Discussion
latitudinal run-up distribution. For the Ecuador–
Colombia and Colombia scenarios, there is also a We modeled in a 30-arc-second grid bathymetry.
longitudinal distribution corresponding to the south- Of course, this is a proxy of finer grids or reality.
ern Panama coastal border. However, the idea here is to present an order of
These distributions are used to calculate marginal magnitude of the run-up in active seismic gaps in the
exceedance probabilities by latitude for each scenario subduction zone of South America. We are trying to
as seen in Fig. 3. In these figures, extreme scenarios model as accurate as is possible with a 30 arc-second
(lighter colors) can be seen with run-ups exceeding grid bathymetry. We are aware that even for a small
20 m and 40 m in some simulations. Nevertheless, area, a port or a bay is mandatory to use fine grids to
the most probable scenarios (darker colors) for run-up account for all the nonlinear effects that are produced
are below 10 m, and in some extreme scenarios in the shoaling process; this would be the most
slightly below 15 or 20 m. Spanning more magni- accurate result that one can obtain, still not the
tudes with this approach would allow hazard curves observed values, especially in the inundation process.
to be constructed by combining marginal probabili-
ties with the occurrence rate of a given earthquake
magnitude in a Poissonian process framework (De 16. Results Validation
Risi & Goda, 2016).
In Fig. 4, the maximum run-up and slip pairs are For validation of the obtained results, we observe
plotted for every simulation in the set of 10 scenarios. Fig. 4. Maximum run-up and slip are in good
A linear least-squares fit is calculated between these agreement with South American and globally recor-
slips and run-ups, and the results are plotted with one ded tsunamis. Furthermore, the Plafker rule of thumb,
standard deviation. Distributions by country are also also known as Plafker law, states that for subduction
included with their corresponding linear fit. earthquakes, maximum run-up should scale with
In general, some results among all scenarios are maximum slip. A constrained null-intercept linear fit
that there is a high-variability zone in front of the between these two parameters shows that run-up
rupture area with run-ups ranging from 10 to 45 m scales with 59–102% of the slip value (slope 1
depending on the earthquake size. Immediately north deviation).
and south there is a low-variability zone where run-
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 4
a Maximum run-up vs. maximum slip for every modeled scenario in circles; fitted linear regression in dotted lines shows that run-up scales up
to 59–102% of the slip for any given scenario. Significant historical subduction tsunamis in South America and the world are denoted by light
and dark symbols, respectively. b Scenarios in Chile, c scenarios in Peru, and d scenarios in Ecuador and Colombia

Tsunami run-up distribution in Peru scales nar- 17. Obtained Run-Up Distributions
rowly with slip at 0.66–0.89 times, a much smaller
range than the Chile and Ecuador–Colombia distri- Run-up distributions show two clear regions: high
butions, both with similar results in the 0.53–0.91 and variability near the rupture, and low variability far
0.60–1.00 ranges, respectively. With the exception of from it. This can be explained because, far from the
the Northern Chile scenario (Mw 9.0), tsunami run- source, magnitude, location, and focal mechanism are
ups in Chile are the ones that scale at the biggest the most important parameters for run-up estimation,
factor with maximum slip. Special care needs to be all of which remain constant between simulations. On
taken when these values are compared, because they the other hand, slip distribution variations have a
belong to different zones with different properties and much more important role near the source (Crempien
generated with a different earthquake. et al., 2020), which explains high variability in run-
up.
With a higher magnitude we also observe higher
variability (see Fig. 4). This can be seen by
M. Medina et al. Pure Appl. Geophys.

comparing the results for a lower-magnitude earth- (2) A bigger rupture zone as a consequence of the
quake (Fig. 3d, e) with a higher-magnitude Arica Bend causes slip distribution to be less densely
earthquake (Fig. 3c, g). For the Northern Chile Up packed as opposed to other scenarios. (3) Nonlinear
scenario (Mw 8.3), at latitude 17.6° S we see the behavior of tsunami waves near the coast might affect
maximum run up of * 18.4 m with 95% of the run- the relationship between run-up and slip. Neverthe-
ups contained in the 7.5–16-m range along with a less, results in this area are similar to those obtained
standard deviation of 2.45 m. In contrast, for the by Ruiz et al. (2015).
Northern Chile scenario (Mw 9.0), we see a * 47-m
maximum value at latitude 20.5° S, with 95% of the
values in the 8–30-m range, having a standard devi- 18. Local and Regional Amplification Phenomena
ation of 6.8 m, about three times as much variability
in the observed run-up. The physical explanation is Although in low-variability zones run-ups are
that with higher magnitude there is more slip to dis- mostly below 5–10 m, there are still some interesting
tribute among simulations, and this redistribution has phenomena in these regions. There are some points
higher consequences in the resulting inundation. on the coast that show constant levels of high run-up
For the run-up in the Northern Chile scenario among all simulations. In Fig. 5 there is an example
(Mw 9.0), run-up values obtained are not in line with of at least two coastal points, around 35° S and 35.5°
other simulations as can be seen in Fig. 4. This can be S, with constant peaks in the run-up value in all the
attributed to several factors, including the following: simulations. We previously defined this constant peak
(1) The geometrical setting in this region, in which as a local amplification phenomenon.
the bend in the subduction at * 18.4° S in front of As seen in Fig. 3g, run-ups in the northern section
Arica, known as the ‘‘Arica Bend,’’ causes the coast of the Central Peru scenario are consistently higher
to be farther from the trench, going from a mean than those obtained south of the source. We define
100-km distance up to 200 km (Saillard et al., 2017). this as a regional amplification phenomenon, when an

Figure 5
Run-up distribution for 50 randomly selected scenarios in South-Central Chile (36°S–29°S) for the North-Central scenario. As shown in the
figure, there is a regular appearance of peaks in run-up among all scenarios at various latitudes
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Table 2
Identified local amplification across all scenarios

Scenario Latitude (°S) f b (°) jbnþ1 j1=2 Nearest location

Central Chile 29.4 4 0.61 9.69 Chungungo


30.0 5 0.73 8.21 Coquimbo
35.3 5 0.17 18.36 Constitución
36.8 3 0.50 10.70 Concepción
North-Central Chile 23.5 4 0.54 10.30 Mejillones Peninsula
29.9 5 1.19 6.35 Coquimbo
30.3 4 4.09 3.74 Tongoy
31.8 6 2.52 4.77 Los Vilos
35.3 4 0.17 18.36 Constitucion
Northern Chile (Up and Down) 16.6 4 0.94 7.81 Camana
18.4 6 0.84 8.26 Arica
23.4 4 5.65 3.18 Mejillones Peninsula
20.2 3 1.29 6.66 Iquique
21.4 4.5 0.91 7.93 Loa River
Southern Peru 11.3 4 0.34 12.98 Huacho Peninsula
12.1 3 0.25 15.14 Chorrillos
16.6 5 0.94 7.81 Camana
18.4 7 0.84 8.26 Arica
Central Peru 15 5 0.34 19.98 Puerto Caballas
Northern Peru 3.6 6 0.41 11.82 Caleta Cruz
3.8 4 0.89 8.02 Cancas
10.7 3.5 0.13 20.99 Pativilca River
11.2 3.5 0.34 19.98 Huacho Peninsula
Locations that repeat across different scenarios are bolded. Amplification factor f is calculated as the peak run-up divided by mean run-up in
the immediate vicinity. Beach slope b is obtained using the last value in a piecewise bathymetry approximation, and slope factor bnþ1 is
calculated as: bnþ1 ¼ tan(b)

irregular run-up is identified over an extended area cases (except for Concepcion), the coast is closer to
([ 100 km), for example, the northern section for the the trench, standing out in comparison to the nearby
Central Peru scenario or the western Panama coast in coast. This could, in fact, mean that these coastal
the Ecuador–Colombia scenario. This definition is in points act as a tsunami barrier stopping wave prop-
contrast to a local amplification which takes place in agation either south or north, and locally increasing
one specific location. run-up values. In particular for the Central Peru
We identified 23 instances of local amplification scenario, the coastal shape at 14° S acts as a barrier,
and two cases of regional amplification, all of which lowering run-up values to the south of this latitude
can be seen in Table 2. We studied the possible (Fig. 6).
causes for this amplification phenomena and identi- The coastal shape in Concepcion has been shown
fied the following: coastal shape, beach slope, to amplify run-up because of shelf resonance during
tsunami directivity, and tsunami clustering. the 2010 Mw 8.8 Maule earthquake (Yamazaki &
Cheung, 2011) and more recently has been shown to
be one of the largest resonators in this area (Aranguiz
19. Coastal Shape et al., 2019).

In this case we identify four instances in which


coast shape could be the cause of the run-up ampli- 20. Directivity
fication: Huacho (11.2° S), Mejillones Peninsula
(23.4° S), Concepcion (36.8° S), and in the regional Directivity amplification is identified in the
case of the Central Peru scenario. In most of these regional case for the Ecuador–Colombia scenario, in
M. Medina et al. Pure Appl. Geophys.

Figure 6
Tsunami maximum wave height showing amplification in a Central Peru scenarios, by a coastal barrier stopping wave propagation southwards
and b Ecuador scenarios, due to tsunami directivity into the west Panama coast

which the western Panama coast has slightly higher with low b values. Fuentes et al. (2015) showed that
values than those observed in the eastern side. In this in a 1 ? 1D beach geometry setting, the slope closest
case, this phenomenon can be explained due to tsu- to the coast (bnþ1 ) plays an important role in the final
nami directivity creating a higher-energy zone maximum run-up value.
directly in front of the source, located on the western To verify this hypothesis in our 2 ? 1D setting,
coast (see Fig. 6). we show b values for every point along the coast
Tsunami energy radiation is similar to that of from Southern Chile (40° S) up to Ecuador at 0°.
Rayleigh waves (Okal, 1988) and has a maximum Examples for low and high b values can be seen in
amplitude in front of the rupture for subduction-type Fig. 7c and d, respectively. Results for all South
earthquakes. This means that there are cases in which American coasts can be seen in Fig. 7e.
the tsunami wave propagation could create higher- As can be seen in Fig. 7a and b, no correlation can
energy zones for the coast located directly in front of be obtained between local amplification value f and b
the rupture area. For example, for an earthquake in or jbnþ1 j1=2 , where bnþ1 ¼ tan(b). Nevertheless,
Central Chile, the tsunami waves could be higher in Fig. 7e shows a good agreement between local
Juan Fernandez Islands than those seen in the north or amplification of run-up R and low or local minimum
south of Chile due to these islands being located b values (below 0.3°). The identified local amplifi-
directly in front of the rupture area as opposed to the cation phenomena can be explained by a local
sides of it. minimum in b value or overall low b values
(b \ 0.3°), with the exception of Puerto Caballas
(15° S) and Iquique (20.2° S) in which no correlation
21. Beach Slope is found.
It is a possibility that the selected set of scenarios
Finally, we study the effect that beach slope b could trigger directivity in the identified coastal
could have in local amplification. As seen in Table 2, points, and the other localities with a low b value
most of the identified local amplifications are in areas could be vulnerable to another set of scenarios.
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 7
Relationship between different beach slope parameters and run-up amplification value. a Amplification value f vs. beach slope b. b
Amplification value f vs. bnþ1 slope: [tan(b)]. Both c and d show examples of a piecewise linear approximation of a bathymetry profile for a
small and large b value, respectively. e Beach slope along latitude b; run-up R amplification marked with dotted lines, and b values lower than
0.3° or local minimum are highlighted with a darker background

22. Tsunami Clustering 23. Possible Identified Causes

Tsunami clustering is defined by stochastic sce- Table 3 shows a summary of all the possible
narios that can form groups with similar inundation identified causes for amplified run-up along the South
patterns (Williamson et al., 2020), directly related to American coast, local or regional. All of these loca-
the earthquake sources. This could be a possible tions could then have an inherently higher
explanation for the local amplification we see in vulnerability to tsunami inundation.
Fig. 5. Nevertheless, several subsets of 50 scenarios In most cases, a low or locally low b value can
were chosen at random with all of them showing explain this feature. There are at least five cases in
similar behaviors for inundation, and given the large which the run-up amplification could be caused by
discrepancies in slip distribution among different other reasons, and only Iquique at 20.2° S remains
stochastic simulations, we do not expect this partic- unexplained with the proposed causes.
ular observed phenomenon to be attributed to tsunami
clustering. Similar tests were done for other
scenarios.
M. Medina et al. Pure Appl. Geophys.

Table 3
Possible identified causes for 17 local and two regional tsunami amplifications

Latitude (°S) Location Possible cause

3.6 Caleta Cruz Beach slope


3.8 Cancas Beach slope
6–9 Northern Peru Apparent amplification due to coastal barrier
10.7 Pativilca River Beach slope
11.2 and 11.3 Huacho Beach slope and coast shape
12.1 Chorrillos Beach slope
15 Puerto Caballas Beach slope
16.6 Camana Beach slope
18.4 Arica Beach slope
20.2 Iquique Non identified
21.4 Loa River Beach slope
23.4 and 23.5 Mejillones Peninsula Beach slope and coast shape
29.4 Chungungo Beach slope
29.9 and 30.0 Coquimbo Beach slope
30.3 Tongoy Beach slope
31.8 Los Vilos Beach slope
35.3 Constitucion Beach slope
36.8 Concepción Beach slope and possible shelf resonance
81.5–84a Costa Rica and Panamá Tsunami directivity
In 16 cases, a low value or local minimum of beach slope can explain tsunami amplification. Causes other than beach slope are bolded. Only
in one case could a cause for local tsunami amplification not be identified
a
Longitude (°W) instead of latitude

24. Conclusions behavior of tsunami run-up (González et al., 2020).


Both studies agree in terms of run-up values, and in
Tsunami run-up shows an immense variability general the Plafker rule is observed in the
along the Nazca-South America and scales differently simulations.
among different segments: Tsunami inundation in Tsunami hazards can still be high even in areas
Chile scales with a greater percentage of maximum where great earthquakes (M [ 8.0) have occurred
slip in comparison to the rest of South America, recently. For instance, our simulations for Central
except for the Northern Chile. This is probably Peru (see Fig. 3g) show that tsunami run-up can
caused by a greater trench-coast distance due to reach 40 m in extreme cases and 15–20 m in most of
change in subduction orientation known as the ‘‘Arica the cases, even though three great earthquakes have
Bend’’ at around 18.4° S. Tsunamis in Peru scale taken place in this region in the last 80 years (1940,
much more narrowly with 63–87% of the maximum 1974, and 2007).
slip. This significant difference is probably associated Even in areas far from the source, where run-up is
with geometric constraints related to the trench-coast usually between one and two meters, local amplifi-
distance being greater in Peru than in Chile and the cation can cause run-up to be much higher than
deformation produced inland due to the tectonic set- expected up to a factor of six or seven times. Local
ting. Tsunamis in Ecuador and Colombia show amplification phenomena were identified in 17
smaller run-ups and much more dispersive behavior instances for coastal locations and are mostly
(see Fig. 4d). This is explained by the tsunami wave explained by local minimum or extremely small
propagation taking place in a more narrow and closed values of beach slope (b \ 0.3°). A detailed bathy-
area with a coast to the north, south, and east of the metry could help to constrain those values of run-up
modeled sources. in those specific locations. This is of special signifi-
It has been also discussed that the near field in cance for tsunami hazard mitigation due to these
Northern Chile can be affected by the nonlinear coastal locations having greater inherent vulnerability
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

to inundation. We do not expect for this phenomenon Acknowledgements


to be related to tsunami clustering, as the used
tsunamigenic sources are widely different from one This work was supported by Fondecyt 1170218.
another.

Appendix
25. Data and Resources
Grid
All data used in this paper came from published
Grids for numerical simulations are constructed
sources listed in the references. Some plots were
according to the Slab 2.0 model (Hayes et al., 2018).
made using the Generic Mapping Tools version 5.4.4
The following Figs. 8, 9, 10, 11, 12, 13, 14, 15, 16,
and 17 show the domains and source area for each
scenario. Profiles A through D show agreement from
north to south of the constructed grid.

Figure 8
Central Perú scenario. White rectangle depicts the deformation area
M. Medina et al. Pure Appl. Geophys.

Figure 9
North-Central Chile scenario. White rectangle depicts the deformation area

Figure 10
Norther Chile (Up) scenario. White rectangle depicts the deformation area
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 11
Colombia scenario. White rectangle depicts the deformation area

Figure 12
Southern Perú scenario. White rectangle depicts the deformation area
M. Medina et al. Pure Appl. Geophys.

Figure 13
Ecuador–Colombia scenario. White rectangle depicts the deformation area

Figure 14
Northern Perú scenario. White rectangle depicts the deformation area
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Figure 15
Central Chile scenario. White rectangle depicts the deformation area

Figure 16
Northern Chile scenario. White rectangle depicts the deformation area
M. Medina et al. Pure Appl. Geophys.

Figure 17
Northern Chile (Down) scenario. White rectangle depicts the deformation area

Publisher’s Note Springer Nature remains neutral Beck, S. L., & Nishenko, S. P. (1990). Variations in the mode of
great earthquake rupture along the central Peru subduction zone.
with regard to jurisdictional claims in published maps
Geophysical Research Letters, 17(11), 1969–1972.
and institutional affiliations. Becker, J. J., Sandwell, D. T., Smith, W. H. F., Braud, J., Binder,
B., Depner, J., & Ladner, R. (2009). Global bathymetry and
elevation data at 30 arc seconds resolution: SRTM30_PLUS.
Marine Geodesy, 32(4), 355–371.
REFERENCES Béjar-Pizarro, M., Socquet, A., Armijo, R., Carrizo, D., Genrich, J.,
& Simons, M. (2013). Andean structural control on interseismic
Abe, K. (1979). Size of great earthquakes of 1837–1974 inferred coupling in the North Chile subduction zone. Nature Geoscience,
from tsunami data. Journal of Geophysical Research: Solid 6(6), 462.
Earth, 84(B4), 1561–1568. Bilek, S. L., & Ruff, L. J. (2002). Analysis of the 23 June 2001 Mw
Aranguiz, R., Catalán, P. A., Cecioni, C., Bellotti, G., Henriquez, = 8.4 Peru underthrusting earthquake and its aftershocks. Geo-
P., & González, J. (2019). Tsunami resonance and spatial pattern physical Research Letters, 29(20), 21–31.
of natural oscillation modes with multiple resonators. Journal of Blaser, L., Krüger, F., Ohrnberger, M., & Scherbaum, F. (2010).
Geophysical Research: Oceans, 124(11), 7797–7816. Scaling relations of earthquake source parameter estimates with
Argus, D. F., & Gordon, R. G. (1991). No-net-rotation model of special focus on subduction environment. Bulletin of the Seis-
current plate velocities incorporating plate motion model mological Society of America, 100(6), 2914–2926.
NUVEL-1. Geophysical Research Letters, 18(11), 2039–2042. Bourgeois, J., Petroff, C., Yeh, H., Titov, V., Synolakis, C. E.,
Barrientos, S. E. (1988). Slip distribution of the 1985 central Chile Benson, B., & Norabuena, E. (1999). Geologie setting, field
earthquake. Tectonophysics, 145(3–4), 225–241. survey and modeling of the Chimbote, Northern Peru, Tsunami
Becerra, A., Sáez, E., Podestá, L., & Leyton, F. (2016). The 2014 of 21 February 1996. In Seismogenic and tsunamigenic processes
earthquake in Iquique, Chile: Comparison between local soil in shallow subduction zones (pp. 513–540). Birkhäuser: Basel.
conditions and observed damage in the cities of Iquique and Alto Bravo, F., Koch, P., Riquelme, S., Fuentes, M., & Campos, J.
Hospicio. Earthquake Spectra, 32(3), 1489–1505. (2019). Slip distribution of the 1985 Valparaı́so earthquake
Becerra, I., Aránguiz, R., González, J., & Benavente, R. (2020). An constrained with seismic and deformation data. Seismological
improvement of tsunami hazard analysis in Central Chile based Research Letters, 90, 1792–1800.
on stochastic rupture scenarios. Coastal Engineering Journal, Carvajal, M., Cisternas, M., & Catalán, P. A. (2017a). Source of the
62(4), 473–488. 1730 Chilean earthquake from historical records: Implications
for the future tsunami hazard on the coast of Metropolitan Chile.
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Journal of Geophysical Research: Solid Earth, 122(5), Fuentes, M., Riquelme, S., Hayes, G., Medina, M., Melgar, D.,
3648–3660. Vargas, G., et al. (2016). A study of the 2015 Mw 8.3 Illapel
Carvajal, M., Cisternas, M., Gubler, A., Catalán, P. A., Winckler, earthquake and tsunami: Numerical and analytical approaches.
P., & Wesson, R. L. (2017b). Reexamination of the magnitudes Pure Applied Geophysics, 173, 1847–1858.
for the 1906 and 1922 Chilean earthquakes using Japanese tsu- Fuentes, M. A., Ruiz, J. A., & Riquelme, S. (2015). The runup on a
nami amplitudes: Implications for source depth constraints. multilinear sloping beach model. Geophysical Journal Interna-
Journal of Geophysical Research: Solid Earth, 122(1), 4–17. tional, 201(2), 915–928.
Catalán, P. A., Aránguiz, R., González, G., Tomita, T., Cienfuegos, Geist, E. L. (2002). Complex earthquake rupture and local tsuna-
R., González, J., & Gubler, A. (2015). The 1 April 2014 Pisagua mis. Journal of Geophysical Research: Solid Earth, 107(B5),
tsunami: Observations and modeling. Geophysical Research ESE-2.
Letters, 42(8), 2918–2925. Geist, E. L., & Dmowska, R. (1999). Local tsunamis and dis-
Chlieh, M., Perfettini, H., Tavera, H., Avouac, J.-P., Remy, D., tributed slip at the source. In J. Sauber & R. Dmowska (Eds.),
Nocquet, J.-M., et al. (2011). Interseismic coupling and seismic Seismogenic and tsunamigenic processes in shallow subduction
potential along the central Andes subduction zone. Journal of zones. Pageoph Topical Volumes. Basel: Birkhäuser.
Geophysical Research: Solid Earth, 116, B12405. Geist, E. L., & Parsons, T. (2006). Probabilistic analysis of tsunami
Cisternas, M., Atwater, B. F., Torrejón, F., Sawai, Y., Machuca, G., hazards. Natural Hazards, 37(3), 277–314.
Lagos, M., Eipert, A., Youlton, C., Salgado, I., Kamataki, T., Goff, J., Witter, R., Terry, J., & Spiske, M. (2020). Palaeotsunamis
Shishikura, M., Rajendran, C. P., Malik, J. K., Rizal, Y., & in the Sino-Pacific region. Earth-Science Reviews, 210, 103352.
Husni, M. (2005). Predecessors of the giant 1960 Chile earth- González, J., González, G., Aránguiz, R., Melgar, D., Zamora, N.,
quake. Nature, 437(7057), 404–407. Shrivastava, M. N., & Cienfuegos, R. (2020). A hybrid deter-
Comte, D., Eisenberg, A., Lorca, E., Pardo, M., Ponce, L., Sarag- ministic and stochastic approach for tsunami hazard assessment
oni, R., & Suárez, G. (1986). The 1985 central Chile earthquake: in Iquique, Chile. Natural Hazards, 100(1), 231–254.
A repeat of previous great earthquakes in the region? Science, Graves, R. W., & Pitarka, A. (2010). Broadband ground-motion
233(4762), 449–453. simulation using a hybrid approach. Bulletin of the Seismological
Comte, D., & Pardo, M. (1991). Reappraisal of great historical Society of America, 100(5A), 2095–2123.
earthquakes in the northern Chile and southern Peru seismic Gutenberg, B., & Richter, C. F. (1944). Frequency of earthquakes
gaps. Natural Hazards, 4(1), 23–44. in California. Bulletin of the Seismological Society of America,
Crempien, J. G. F., Urrutia, A., Benavente, R., & Cienfuegos, R. 34(4), 185–188.
(2020). Effects of earthquake spatial slip correlation on vari- Hayes, G. P., Herman, M. W., Barnhart, W. D., Furlong, K. P.,
ability of tsunami potential energy and intensities. Scientific Riquelme, S., Benz, H. M., & Samsonov, S. (2014). Continuing
Reports, 10, 8399. megathrust earthquake potential in Chile after the 2014 Iquique
De Risi, R., & Goda, K. (2016). Probabilistic earthquake–tsunami earthquake. Nature, 512(7514), 295.
multi-hazard analysis: Application to the Tohoku region, Japan. Hayes, G. P., Moore, G. L., Portner, D. E., Hearne, M., Flamme,
Frontiers in Built Environment, 2, 25. H., Furtney, M., & Smoczyk, G. M. (2018). Slab2, a compre-
Delouis, B., Monfret, T., Dorbath, L., Pardo, M., Rivera, L., Comte, hensive subduction zone geometry model. Science, 362(6410),
D., & Cisternas, A. (1997). The Mw = 8.0 Antofagasta (northern 58–61.
Chile) earthquake of 30 July 1995: A precursor to the end of the Herrero, A., & Bernard, P. (1994). A kinematic self-similar rupture
large 1877 gap. Bulletin of the Seismological Society of America, process for earthquakes. Bulletin of the Seismological Society of
87(2), 427–445. America, 84(4), 1216–1228.
Delouis, B., Nocquet, J.-M., & Vallée, M. (2010). Slip distribution Kanamori, H. (1977). The energy release in great earthquakes.
of the February 27, 2010 Mw = 8.8 Maule earthquake, central Journal of Geophysical Research, 82(20), 2981–2987.
Chile, from static and high-rate GPS, InSAR, and broadband Kanamori, H., & Cipar, J. J. (1974). Focal process of the great
teleseismic data. Geophysical Research Letters, 37, L17305. Chilean earthquake May 22, 1960. Physics of the Earth and
Dorbath, L., Cisternas, A., & Dorbath, C. (1990). Assessment of Planetary Interiors, 9(2), 128–136.
the size of large and great historical earthquakes in Peru. Bulletin Kanamori, H., & McNally, K. C. (1982). Variable rupture mode of
of the Seismological Society of America, 80(3), 551–576. the subduction zone along the Ecuador–Colombia coast. Bulletin
Drápela, J., Calisto, I., & Moreno, M. (2021). Locking-derived of the Seismological Society of America, 72(4), 1241–1253.
tsunami scenarios for the most recent megathrust earthquakes in Kendrick, E. C., Bevis, M., Smalley, R. F., Jr., Cifuentes, O., &
Chile: Implications for tsunami hazard assessment. Natural Galban, F. (1999). Current rates of convergence across the cen-
Hazards, 107, 35–52. tral Andes: Estimates from continuous GPS observations.
Dziewonski, A. M., & Anderson, D. L. (1981). Preliminary refer- Geophysical Research Letters, 26(5), 541–544.
ence Earth model. Physics of the Earth and Planetary Interiors, Lay, T., Kanamori, H., Ammon, C. J., Koper, K. D., Hutko, A. R.,
25(4), 297–356. Ye, L., et al. (2012). Depth-varying rupture properties of sub-
Escobar, R. S., Diaz, L. O., Guerrero, A. M., Galindo, M. P., Mas, duction zone megathrust faults. Journal of Geophysical
E., Koshimura, S., & Quintero, P. (2020). Tsunami hazard Research: Solid Earth, 117, B04311.
assessment for the central and southern pacific coast of Colom- Lomnitz, C. (2004). Major earthquakes of Chile: A historical sur-
bia. Coastal Engineering Journal, 62(4), 540–552. vey, 1535–1960. Seismological Research Letters, 75(3),
Fuentes, M. S., Riquelme Muñoz, S., Medina, M., Mocanu, M., & 368–378.
Filippi Fernandez, R. (2019). Tsunami hazard evaluation in the Mai, P. M., & Beroza, G. C. (2002). A spatial random field model
Coquimbo region using nonuniform slip distribution sources. to characterize complexity in earthquake slip. Journal of Geo-
Seismological Research Letters, 90, 1812–1819. physical Research: Solid Earth, 107(B11), ESE-10.
M. Medina et al. Pure Appl. Geophys.

Mas, E., Adriano, B., Pulido, N., Jimenez, C., & Koshimura, S. Ruiz, S., Klein, E., del Campo, F., Rivera, E., Poli, P., Metois, M.,
(2014). Simulation of tsunami inundation in Central Peru from & Madariaga, R. (2016). The seismic sequence of the 16
future megathrust earthquake scenarios. Journal of Disaster September 2015 M w 8.3 Illapel, Chile, earthquake. Seismolog-
Research, 9(6), 961–967. ical Research Letters, 87(4), 789–799.
McCann, W. R., Nishenko, S. P., Sykes, L. R., & Krause, J. (1979). Saillard, M., Audin, L., Rousset, B., Avouac, J. P., Chlieh, M.,
Seismic gaps and plate tectonics: Seismic potential for major Hall, S. R., & Farber, D. L. (2017). From the seismic cycle to
boundaries. In M. Wyss (Ed.), Earthquake prediction and seis- long-term deformation: Linking seismic coupling and Quaternary
micity patterns. Contributions to Current Research in coastal geomorphology along the Andean megathrust. Tectonics,
Geophysics. Basel: Birkhäuser. 36(2), 241–256.
Melgar, D., & Hayes, G. P. (2019). The correlation lengths and Scholz, C. H., & Campos, J. (1995). On the mechanism of seismic
hypocentral positions of great earthquakes. Bulletin of the Seis- decoupling and back arc spreading at subduction zones. Journal
mological Society of America, 109(6), 2582–2593. of Geophysical Research: Solid Earth, 100(B11), 22103–22115.
Métois, M., Socquet, A., & Vigny, C. (2012). Interseismic cou- Scholz, C. H., & Campos, J. (2012). The seismic coupling of
pling, segmentation and mechanical behavior of the central Chile subduction zones revisited. Journal of Geophysical Research:
subduction zone. Journal of Geophysical Research: Solid Earth, Solid Earth, 117, B05310.
117, B03406. Sieh, K., Natawidjaja, D. H., Meltzner, A. J., Shen, C. C., Cheng,
Métois, M., Socquet, A., Vigny, C., Carrizo, D., Peyrat, S., H., Li, K. S., & Edwards, R. L. (2008). Earthquake supercycles
Delorme, A., & Ortega, I. (2013). Revisiting the North Chile inferred from sea-level changes recorded in the corals of west
seismic gap segmentation using GPS-derived interseismic cou- Sumatra. Science, 322(5908), 1674–1678.
pling. Geophysical Journal International, 194(3), 1283–1294. Sladen, A., Tavera, H., Simons, M., Avouac, J. P., Konca, A. O.,
Métois, M., Vigny, C., & Socquet, A. (2016). Interseismic cou- Perfettini, H., et al. (2010). Source model of the 2007 Mw 8.0
pling, megathrust earthquakes and seismic swarms along the Pisco, Peru earthquake: Implications for seismogenic behavior of
Chilean subduction zone (38–18 S). Pure and Applied Geo- subduction megathrusts. Journal of Geophysical Research: Solid
physics, 173(5), 1431–1449. Earth, 115, B02405.
Moreno, M. S., Bolte, J., Klotz, J., & Melnick, D. (2009). Impact of Spiske, M., Piepenbreier, J., Benavente, C., Kunz, A., Bahlburg, H.,
megathrust geometry on inversion of coseismic slip from & Steffahn, J. (2013). Historical tsunami deposits in Peru: Sed-
geodetic data: Application to the 1960 Chile earthquake. Geo- imentology, inverse modeling and optically stimulated
physical Research Letters, 36, L16310. luminescence dating. Quaternary International, 305, 31–44.
Mothes, P. A., Rolandone, F., Nocquet, J. M., Jarrin, P. A., Strasser, F. O., Arango, M. C., & Bommer, J. J. (2010). Scaling of
Alvarado, A. P., Ruiz, M. C., & Segovia, M. (2018). Monitoring the source dimensions of interface and intraslab subduction-zone
the earthquake cycle in the northern Andes from the Ecuadorian earthquakes with moment magnitude. Seismological Research
cGPS network. Seismological Research Letters, 89(2A), Letters, 81(6), 941–950.
534–541. Tichelaar, B. W., & Ruff, L. J. (1993). Depth of seismic coupling
Nocquet, J. M., Jarrin, P., Vallée, M., Mothes, P. A., Grandin, R., along subduction zones. Journal of Geophysical Research: Solid
Rolandone, F., & Régnier, M. (2017). Supercycle at the Earth, 98(B2), 2017–2037.
Ecuadorian subduction zone revealed after the 2016 Pedernales Trenkamp, R., Kellogg, J. N., Freymueller, J. T., & Mora, H. P.
earthquake. Nature Geoscience, 10(2), 145. (2002). Wide plate margin deformation, southern Central
Nocquet, J. M., Villegas-Lanza, J. C., Chlieh, M., Mothes, P. A., America and northwestern South America, CASA GPS obser-
Rolandone, F., Jarrin, P., & Martin, X. (2014). Motion of con- vations. Journal of South American Earth Sciences, 15(2),
tinental slivers and creeping subduction in the northern Andes. 157–171.
Nature Geoscience, 7(4), 287. Udı́as, A., Madariaga, R., Buforn, E., Muñoz, D., & Ros, M.
Okada, Y. (1985). Surface deformation due to shear and tensile (2012). The large Chilean historical earthquakes of 1647, 1657,
faults in a half-space. Bulletin of the Seismological Society of 1730, and 1751 from contemporary documents. Bulletin of the
America, 75(4), 1135–1154. Seismological Society of America, 102(4), 1639–1653.
Okal, E. A. (1988). Seismic parameters controlling far-field tsu- Vargas, C. A., Caneva, A., Monsalve, H., Salcedo, E., & Mora, H.
nami amplitudes: A review. Natural Hazards, 1(1), 67–96. (2018). Geophysical networks in Colombia. Seismological
Okal, E. A., Borrero, J. C., & Synolakis, C. E. (2006). Evaluation Research Letters, 89(2A), 440–445.
of tsunami risk from regional earthquakes at Pisco, Peru. Bulletin Vigny, C., Rudloff, A., Ruegg, J. C., Madariaga, R., Campos, J., &
of the Seismological Society of America, 96(5), 1634–1648. Alvarez, M. (2009). Upper plate deformation measured by GPS
Pacheco, J. F., Sykes, L. R., & Scholz, C. H. (1993). Nature of in the Coquimbo Gap, Chile. Physics of the Earth and Planetary
seismic coupling along simple plate boundaries of the subduction Interiors, 175(1–2), 86–95.
type. Journal of Geophysical Research: Solid Earth, 98(B8), Vigny, C., Socquet, A., Peyrat, S., Ruegg, J. C., Métois, M.,
14133–14159. Madariaga, R., & Carrizo, D. (2011). The 2010 Mw 8.8 Maule
Pelayo, A. M., & Wiens, D. A. (1990). The November 20, 1960 megathrust earthquake of central Chile, monitored by GPS.
Peru tsunami earthquake: Source mechanism of a slow event. Science, 332(6036), 1417–1421.
Geophysical Research Letters, 17(6), 661–664. Villegas-Lanza, J. C., Chlieh, M., Cavalié, O., Tavera, H., Baby, P.,
Ruiz, J. A., Fuentes, M., Riquelme, S., Campos, J., & Cisternas, A. Chire-Chira, J., & Nocquet, J. M. (2016a). Active tectonics of
(2015). Numerical simulation of tsunami runup in northern Chile Peru: Heterogeneous interseismic coupling along the Nazca
based on non-uniform k-2 slip distributions. Natural Hazards, megathrust, rigid motion of the Peruvian Sliver, and Subandean
79(2), 1177–1198. shortening accommodation. Journal of Geophysical Research:
Solid Earth, 121(10), 7371–7394.
Tsunami Modeling in the South American Subduction Zone Inferred from Seismic Coupling

Villegas-Lanza, J. C., Nocquet, J. M., Rolandone, F., Vallée, M., Yamazaki, Y., Kowalik, Z., & Cheung, K. F. (2009). Depth-inte-
Tavera, H., Bondoux, F., & Chlieh, M. (2016b). A mixed seis- grated, non-hydrostatic model for wave breaking and run-up.
mic–aseismic stress release episode in the Andean subduction International Journal for Numerical Methods in Fluids, 61(5),
zone. Nature Geoscience, 9(2), 150–154. 473–497.
Williamson, A., Rim, D., Adams, L., LeVeque, R. J., Melgar, D., & Ye, L., Kanamori, H., Avouac, J. P., Li, L., Cheung, K. F., & Lay,
Gonzalez, F. (2020). A source clustering approach for efficient T. (2016). The 16 April 2016, MW 7.8 (MS 7.5) Ecuador
inundation modeling and regional scale PTHA earthquake: A quasi-repeat of the 1942 MS 7.5 earthquake and
Yamazaki, Y., & Cheung, K. F. (2011). Shelf resonance and impact partial re-rupture of the 1906 MS 8.6 Colombia-Ecuador earth-
of near-field tsunami generated by the 2010 Chile earthquake. quake. Earth and Planetary Science Letters, 454, 248–258.
Geophysical Research Letters, 38, L12605.

(Received January 14, 2021, revised June 23, 2021, accepted June 24, 2021)

You might also like