Depth-dependent rupture mode along the Ecuador-Colombia subduction zone: Depth-dependent rupture mode

Abstract

A large earthquake (Mw 7.7) occurred on 16 April 2016 within the source region of the 1906 earthquake in the Ecuador-Colombia subduction zone. The 1906 event has been interpreted as a megathrust earthquake (Mw 8.8) that ruptured the source regions of smaller earthquakes in 1942, 1958, and 1979 in this subduction. Our seismic analysis indicated that the spatial distribution of the 2016 earthquake and its aftershocks correlated with patches of high interplate coupling strength and was similar to those of the 1942 earthquake and its aftershocks, suggesting that the 2016 and 1942 earthquakes ruptured the same asperity. Our analysis of tsunami waveforms of the 1906 event indicated Mw around 8.4 and showed that large slip occurred near the trench off the source regions of the above three historical and the 2016 earthquakes, suggesting that a depth-dependent complex rupture mode exists along this subduction zone.

Full text

Depth-dependent rupture mode along the Ecuador-
Colombia subduction zone
Masahiro Yoshimoto
1
, Hiroyuki Kumagai
1
, Wilson Acero
2
, Gabriela Ponce
2
,
Freddy Vásconez
2
, Santiago Arrais
2
, Mario Ruiz
2
, Alexandra Alvarado
2
,
Patricia Pedraza García
3
, Viviana Dionicio
3
, Orlando Chamorro
3
, Yuta Maeda
1
,
and Masaru Nakano
4
1
Graduate School of Environmental Studies, Nagoya University, Nagoya, Japan,
2
Instituto Geofísico, Escuela Politécnica
Nacional, Quito, Ecuador,
3
Servicio Geológico Colombiano, Bogota, Colombia,
4
Japan Agency for Marine-Earth Science and
Technology, Yokohama, Japan
Abstract A large earthquake (M
w
7.7) occurred on 16 April 2016 within the source region of the 1906
earthquake in the Ecuador-Colombia subduction zone. The 1906 event has been interpreted as a
megathrust earthquake (M
w
8.8) that ruptured the source regions of smaller earthquakes in 1942, 1958, and
1979 in this subduction. Our seismic analysis indicated that the spatial distribution of the 2016 earthquake
and its aftershocks correlated with patches of high interplate coupling strength and was similar to those of
the 1942 earthquake and its aftershocks, suggesting that the 2016 and 1942 earthquakes ruptured the same
asperity. Our analysis of tsunami waveforms of the 1906 event indicated M
w
around 8.4 and showed that
large slip occurred near the trench off the source regions of the above three historical and the 2016
earthquakes, suggesting that a depth-dependent complex rupture mode exists along this subduction zone.
1. Introduction
Studies in the Ecuador-Colombia subduction zone, where the Nazca plate is subducting beneath the
South American plate, have made important contributions to our understanding of the role played by
asperities in interplate earthquakes. Along this subduction zone, three large earthquakes occurred in
1942 (M
w
7.8) [Swenson and Beck, 1996], 1958 (M
w
7.7) [Kanamori and McNally, 1982], and 1979 (M
w
8.1)
(Global CMT Project, http://www.globalcmt.org/) next to each other. Further, an earthquake that occurred
in 1906 has been interpreted as a megathrust event with M
w
8.8 [Kanamori and McNally, 1982] on
the basis of its area of strong ground motion and ground deformation [Kelleher, 1972] and of far-field
tsunami heights [Abe, 1979]. Kanamori and McNally [1982] considered that the 1906 earthquake ruptured
asperities of the above three earthquakes simultaneously. On the basis of this rupture history, they
proposed that the magnitude of a megathrust earthquake that ruptures multiple asperities of smaller
earthquakes could grow larger than expected sum of the magnitudes of smaller events that rupture indi-
vidual asperities.
However, Okal [1992], who analyzed long-period surface wave records, suggested that the 1906 earthquake
was not as large as M
w
8.8 and did not exceed M
w
8.5. Tsuzuki et al. [2012] reevaluated tsunami heights
associated with the 1906 earthquake and suggested that the earthquake could be explained by an M
w
8.5
event corresponding to failure of two source regions of the 1958 and 1979 earthquakes. These results are
inconsistent with the interpretation of Kanamori and McNally [1982]. Since the 1906 earthquake is the largest
event ever recorded in this subduction zone, it is important to estimate the actual size of the event for the
evaluation of earthquake hazards.
Various studies have investigated the seismicity [Font et al., 2013], structure [Collot et al., 2004; Gailler et al.,
2007], deformation [White et al., 2003; Chlieh et al., 2014], and slow-slip events [Vallée et al., 2013] along this
subduction zone. The Instituto Geofísico, Escuela Politécnica Nacional (IG-EPN), the earthquake and volcano
monitoring institute of Ecuador, has upgraded the national seismic network with broadband seismometers.
In addition, it has recently become possible to accurately calculate tsunami waveforms at trans-Pacific
distances by taking into account the dispersion effects caused by elastic tsunami loadings on the solid earth,
compression and dilatation of seawater, and gravitational potential change associated with mass motion
during tsunami propagation [e.g., Watada et al., 2014; Yoshimoto et al., 2016].
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2203
PUBLICATION
S
Geophysical Research Letters
RESEARCH LETTER
10.1002/2016GL071929
Key Points:
•The 2016 Ecuador earthquake was
analyzed by using regional and global
broadband seismic data
•Source slip inversion of the 1906
earthquake was performed by using
far-field tsunami waveforms
•Large slip of the 1906 earthquake
(M
w
8.4) occurred near the trench off
the source regions of the 2016 and
other historical earthquakes
Supporting Information:
•Supporting Information S1
Correspondence to:
M. Yoshimoto,
yoshimoto.masahiro@a.mbox.nagoya-
u.ac.jp
Citation:
Yoshimoto, M., et al. (2017),
Depth-dependent rupture mode
along the Ecuador-Colombia
subduction zone, Geophys. Res. Lett.,
44, 2203–2210, doi:10.1002/
2016GL071929.
Received 4 AUG 2016
Accepted 22 FEB 2017
Accepted article online 25 FEB 2017
Published online 11 MAR 2017
©2017. American Geophysical Union.
All Rights Reserved.

In this paper, we studied the seismic sources of the 2016 earthquake and its aftershocks and the tsunami
source of the 1906 earthquake. Using our source estimates, we investigated the 2016 earthquake in relation
to the rupture history of earthquakes in the Ecuador-Colombia subduction zone.
2. Methods and Data
2.1. Centroid Moment Tensor Inversion
To estimate centroid moment tensor (CMT) mechanisms from regional broadband seismic data, we used the
SWIFT method [Nakano et al., 2008; Sakai et al., 2016], in which waveform inversion is performed in the
frequency domain to estimate the moment function by assuming a point source and a double-couple
mechanism. In this method, a spatial grid search of three fault angles (strike, dip, and rake) is made to find
the best fitting fault parameters and the source centroid that minimizes the residuals between observed
and synthetic displacement seismograms (see Text S1 in the supporting information). We used Green’s
functions calculated by the discrete wave number method [e.g., Bouchon, 1981] and the standard ak135
Earth model [Kennett et al., 1995]. We followed the fault angle definition of Aki and Richards [2002] through-
out this study.
Broadband seismic data used in this study were from the IG-EPN (Ecuador) and Servicio Geológico
Colombiano (SGC) (Colombia) regional seismic networks. We used seismic data from 27 stations with broad-
band seismometers (Nanometrics Trillium 120 and Compact) and 3 stations with velocity-type strong-motion
broadband seismometers (Tokyo Keiki TSM-1) in the Ecuador network (Figure 1a). The velocity-type strong-
motion broadband seismometers having a measurable range of ±3 m/s were deployed in February–March
2016 just before the occurrence of the 2016 earthquake. We also used seismic data from 11 stations with
ordinary broadband seismometers in the Colombia network (Figure S1a).
2.2. Inversion of Telseismic Data
To perform the inversion of teleseismic data of the 2016 earthquake, we calculated Green’s functions using
Earth model IASP91 [Kennett and Engdahl, 1991] by the method of Kikuchi and Kanamori [1991] and applied
the waveform inversion scheme of Kikuchi et al. [2003]. We used a fault plane subdivided into subfaults with
dimensions of 20 km × 20 km. A slip rate function on each subfault composed of 10 triangular functions with a
duration of 4 s at every 2 s. We used a constant rupture velocity of 3000 m/s and a rigidity of 40 GPa. We used
vertical components of teleseismic Pwave records band passed between 1 and 250 s at 54 stations at epicen-
tral distances between 30° and 100° from the epicenter of the 2016 event. We performed the inversion of
these data to estimate spatiotemporal slip distributions, in which a grid search in strike and dip angles of
the fault plane was conducted to find the best fitting fault orientation.
2.3. Tsunami Waveform Analysis
For our tsunami analysis, we first developed seven uniform slip models for M
w
between 8.2 and 8.8 to
estimate the size of the 1906 earthquake based on the simple tsunami forward modeling. In each model,
we assumed the centroid location at the center of the source area estimated by Kanamori and McNally
[1982]; we applied the fault orientation of the Global CMT solution for the 1979 earthquake (30°, 16°, and
118°) and the scaling relation of Murotani et al. [2013] to construct the models. The fault plane spread from
the ocean floor along the trench to the depth of the maximum fault width of 160 km at most. In case the fault
width exceeded 160 km, we extended the fault length following the scaling relation. We simulated tsunami
waveforms using these models, and then comparison was made of the simulated waveforms to observed
tsunami records. A following normalized residual was calculated:
R¼
1
NX
N
i¼1
Ao
iAc
i

2
Ao
i

2;(1)
where Ao
iand Ac
iare root-mean-square amplitudes of observed and simulated waveforms, respectively, and N
is the number of stations. We then constructed a fault model for our tsunami inversion. We subdivided the
fault plane for M
w
8.8 into 3 × 11 subfaults along the dip and strike directions, respectively, in which each sub-
fault had a 50 km× 50km area. We assumed strike and rake angles of 30° and 118° from the Global CMT solu-
tion of the 1979 event, respectively, for all the subfaults. In accordance with slab model Slab 1.0 [Hayes et al.,
2012], we used dip angles of 14°, 22°, and 26° for shallow, intermediate, and deep subfaults, respectively.
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2204

We calculated the displacement of the ocean floor due to each subfault dislocation using the analytical equa-
tions of Okada [1985] and taking into account the effect of coseismic horizontal displacement in regions of
steep bathymetric slopes [Tanioka and Satake, 1996]. We simulated linear-long tsunami waves from each
subfault using the finite-difference method of Fujii and Satake [2007] with a constant rise time of 30 s. We
corrected the phase of the simulated tsunami waves by the method of Watada et al. [2014] to include the
dispersion effects. We followed the slip inversion scheme of Fujii and Satake [2013] and used 15:35 on 31
January 1906 (UTC) as the origin time to estimate slip on each subfault.
We digitized the tide gauge records of the 1906 earthquake at Honolulu (Hawaii), San Francisco (California),
and Ayukawa (Japan) with a sampling interval of 1 min. We note that clear tsunami signals were visible only in
Figure 1. (a) Locations of broadband seismic stations. Red triangles represent stations with velocity-type strong-motion
seismometers (Tokyo Keiki TSM-1). Each station indicated by a green triangle features either a Nanometrics Trillium 120
or Compact broadband seismometer. The rectangle indicates the area shown in Figures 1b–1d. (b) Centroid moment
tensor mechanisms of the 2016 earthquake (M
w
7.7) and aftershocks with M
w
>4.5. Dots indicate source locations of
aftershocks with magnitudes larger than 4.0. Contour lines with an interval of 0.4 m represent the slip distribution
estimated by teleseismic waveform inversion (see Figure S3). (c) Interseismic plate coupling estimated by Chlieh et al.
[2014]. (d) Source mechanism of the 1942 earthquake (M
w
7.8) and epicenters of aftershocks (black circles) estimated by
Mendoza and Dewey [1984] with quality A or B (epicenter errors within 20 km).
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2205

the record at Honolulu; the other station records were noisy, but useful to constrain the earthquake size. We
removed tidal signals by fitting polynomial functions and calculated the three-point moving average to
suppress short-period noise.
3. Results
3.1. The 2016 Earthquake and Aftershocks
We estimated the CMT of the 2016 event using the SWIFT method. Since all waveform data recorded by
ordinary broadband sensors in the Ecuador network were saturated, we used those by velocity-type
strong-motion seismometers (Figures 1a and S1a). The inversion results indicated the earthquake was an
M
w
7.7 event with a dip-slip mechanism; the source centroid was located 50 km south of the hypocenter at
a depth of 25 km (Figures 1b and S1b). This solution is consistent with those estimated by using waveform
data from the Colombian network and by the Global CMT Project (Figure S1b). We also estimated the CMT
mechanisms of aftershocks (M
w
>4.5) by using waveform data recorded by the ordinary broadband sensors
in the Ecuador and Colombia networks. The estimated CMT solutions indicate dominantly dip-slip mechan-
isms. Large aftershocks of the 2016 event were found to locate in the north and south of the mainshock slip
region at shallow depths (Figure 1b and Table S1).
To perform the waveform inversion of the teleseismic data to estimate slip distribution of the 2016 earth-
quake, we first used the epicenter determined by the IG-EPN (0.3474°, 80.1577°) and a depth of 19 km for
the rupture initiation point. However, the centroid location estimated from the inversion was inconsistent
with that of the SWIFT CMT solution (Figure S2). The absolute location of the slip distribution is strongly
dependent on the rupture initiation point, but the hypocenter and the rupture initiation point may not
always coincide, especially if the initial phase is slow. The observed waveforms of the 2016 earthquake with
emergent onsets suggest that this is the case. We therefore performed our slip inversion by assuming that the
centroid location of the slip distribution coincided with that of the SWIFT CMT solution. Our slip inversion
indicated that the rupture propagated from north to south and that the maximum slip was 2.2 m (Figures
S3 and S4).
3.2. Tsunami Inversion of the 1906 Earthquake
We first used the seven uniform slip models to estimate the size of the 1906 earthquake. We simulated the
tsunami waveforms using these models and compared the simulated waveforms with the observed records
at the three stations (Figure 2a) within the time range shown by blue bars in Figure 2b. This time range, which
was also used in our tsunami inversion, represents the time interval that includes the first tsunami waves gen-
erated from the 1906 fault model (Figure 3). We used this time range to exclude later waves reflected from
shore near each tide gauge station. As shown in Figure 2b, the slip model with M
w
8.8 overestimated the tsu-
nami amplitudes. The normalized residual Rduring the time range was minimized at M
w
8.4 (Figure 2c).
We next conducted a checkerboard test to estimate the tsunami inversion resolution. We synthesized
tsunami waveforms using a checkerboard slip model with slip of either 0 or 10 m on each subfault (Figure
S5). We then added noise to the synthesized tsunami waveforms using the typical noise level in each tsunami
record. Our checkerboard test inversion showed that the checkerboard patterns were fairly well reproduced
for the shallower subfaults (Figure S5).
The slip distributions which gave the best waveform fits are shown in Figure 3. Our inversion indicated that
the dominant slip occurred in the southern shallow part of the region (Figure 3), where the resolution was
better as shown by the checkerboard test result (Figure S5). The estimated M
w
was 8.4, which is consistent
with the uniform slip model estimate. We compared the observed tsunami record at Honolulu with the
tsunami waveform generated from each subfault (Figure S6). This comparison clearly indicated that the
arrival times of the tsunami waveforms from the southern shallow subfaults (patch 1 in Figure S6) were
consistent with the Honolulu record, supporting our slip inversion results.
Since we used the old tide gauge records, there is the possibility that the clock used in each record was not
accurate. Such time uncertainty may cause an apparent spatial shift of the estimated larger slip area. To
examine this point, we performed the inversion using the Honolulu record with a uniform time shift between
2 min before and 8 min after the original time at an interval of 1 min and other records without the time shift.
The minimum residual between the observed and synthetic waveforms in the inversion was obtained when
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2206

using the Honolulu record without the time shift, and the inversion result showing the dominant slip in the
southern shallow part was stable against the time shift within a few minutes (Figure S7). To further verify the
reliability of the large slip in the shallow part, we removed all the shallowest subfaults from our fault model
and performed the inversion with the time shift for the Honolulu record. The inversion results showed that
the residual values for this partial fault model were clearly larger than those in the original fault model
(Figure S8), indicating that downdip slip may not be able to explain the observed waveforms. These results
support that the large slip of the 1906 earthquake occurred in the southern shallow part near the trench.
We note that the tide gauge record at Naos Island in Panama presented in Soloviev and Go [1975] was not
used in this study. This is because the tsunami inversion including this record provided an unrealistic
Figure 2. (a) Locations of tide gauge stations and distributions of uniform slip models. (b) Comparison of the observed
tsunami records (black lines) with tsunami waveforms (red lines) simulated using the uniform slip models. Time is
measured from the origin time of the 1906 earthquake. (c) Plots of the residual values calculated with equation (1) within
blue bars in Figure 2b against the moment magnitudes of the uniform slip models.
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2207

solution showing anomalously large
localized slip (~30 m) in the northern
and southern areas of the fault model
(Figure S9).
4. Discussion and Conclusions
We compared the CMT and teleseismic
waveform inversion results with an
interseismic plate coupling model (the
rough model of Chlieh et al. [2014]).
Areas of strong interplate coupling
occur as isolated patches, designated
C1 through C6, along the Ecuadorian
coast (Figure 1c). The slip area of the
2016 earthquake overlaps C3
(Figure 4), indicating that the 2016
event ruptured this asperity. The after-
shocks were distributed in and around
patches C1, C2, and C4 (Figure 4).
We next compared the epicenter and
focal mechanism of the 1942 earth-
quake (M
w
7.8) and the epicenter of its
aftershocks [Mendoza and Dewey, 1984]
(Figure 1d) with those of the 2016 event
and its aftershocks (Figure 1b). We
found that the mechanisms, magni-
tudes, and aftershock distributions were
similar to each other between the 2016
and 1942 earthquakes, except for some
aftershocks of the 2016 event that
occurred around 1°S near C1. The
maximum coupling in patch C3, which
corresponds to the large slip region of
the 2016 earthquake, was around 0.8
(Figure 1c). Chlieh et al. [2014] presented
coupling models depending on the
smoothness factor and the moment
deficit rate, and the maximum coupling
in C3 ranged from 0.5 to 0.8 in these
models. If we assume that the conver-
gence rate of the Nazca plate with
respect to the North Andean Sliver
[Nocquet et al., 2014] is 46 mm/yr
[Chlieh et al., 2014], then the estimated
maximum slip deficit accumulated
between 1942 and 2016 at C3 ranges
from 1.7 to 2.7 m; this range is consis-
tent with our estimated maximum slip
of 2.2 m for the 2016 earthquake.
These findings suggest that the 2016 earthquake ruptured the same asperity as the 1942 earthquake and
indicate an earthquake recurrence interval of 74 years. If this asperity was ruptured by the 1906 earthquake
as proposed by Kanamori and McNally [1982], then the 1942 earthquake occurred only 36 years after the 1906
earthquake. This time period is roughly half of the time interval between the 1942 and 2016 events.
Figure 3. Slip distribution estimated from our tsunami waveform inver-
sion of the 1906 earthquake and waveform fits between the observed
(black lines) and synthetic (red lines) tsunami waveforms. The inversion
was performed using the waveforms within blue bars. Time is measured
from the origin time of the 1906 earthquake.
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2208

Our tsunami inversion results indicated
that the main slip of the 1906 earth-
quake occurred near the trench offshore
of the 1942, 1958, and 1979 source
regions (Figure 4), which is different
from the source model proposed by
Kanamori and McNally [1982]. Since our
tsunami inversion resolution in the
intermediate and deep subfaults was
limited (Figure S5), we may not be able
to rule out the possibility of slip in these
regions. However, our estimated M
w
around 8.4 was consistent with the esti-
mate (M
w
<8.5) from surface wave
records [Okal, 1992]. This supports that
the slip near the trench was dominant
compared to that in the deeper sub-
faults. If the 1906 earthquake did not
rupture the 1942 source region, the
discrepancy in the recurrence interval
can be reasonably explained.
Our results indicate that the large slip
region of the 1906 earthquake did not
overlap those of the 1942, 1958, 1979,
and 2016 earthquakes; therefore, there
must be two rupture domains along
the Ecuador-Colombia subduction zone:
one near the trench and the other near
the coast. The trenchward domain
produced the 1906 earthquake with
M
w
8.4, and individual segments in the
coastal domain produced M
w
7 class
earthquakes. This feature points to
the existence of a depth-dependent
complex rupture mode along the
Ecuador-Colombia subduction zone.
Hashimoto et al. [2009] demonstrated
that patches where the coupling
strength is high along the Kuril-Japan
trench correspond to the source regions
of historical and recent large earth-
quakes. Moreno et al. [2010] showed
that the seismic slip associated with
the 2010 Maule megathrust earthquake (M
w
8.8) in Chile correlated with the preseismic interplate coupling
strength. Our study of a M
w
7 class earthquake along the Ecuador-Colombia subduction zone provides further
evidence supporting the hypothesis that preseismic coupling region and coseismic slip area are similar.
Taken together, these findings suggest that this hypothesis may generally hold in subduction zones over a
wide range of seismic magnitudes, and they imply that the distribution of coupling strength can be used
to assess future slip. We consider this possibility here by focusing on the source region of the 1958
earthquake (Figure 4). The slip deficit that has accumulated in C5 since 1958 corresponds to a moment
magnitude of 7.6. Patch C5 is located beneath the ocean floor, and a future event might generate not only
large ground motion but also tsunamis, although apparently no serious tsunami damage was caused by
the 1958 event.
Figure 4. Comparison of the slip distribution of the 1906 earthquake with
the main shocks and aftershocks of the 1958, 1979, and 2016 earthquakes.
The slip areas of the 1906 earthquake are displayed by colored squares
with slip magnitudes indicated by the color scale. Red contours represent
the slip distribution of the 2016 earthquake. Source mechanisms colored
red indicate those of the 2016 earthquake (M
w
7.7) and its aftershocks.
The pink source mechanism is the Global CMT solution of the 1979
earthquake (M
w
8.1), and pink stars indicate the epicenters of the 1979
aftershocks. The source mechanism of the 1958 earthquake (M
w
7.7)
(green) is assumed to be the same as that of the 1979 earthquake, and
green stars indicate the epicenters of the 1958 aftershocks. The epicenters
of the 1979 and 1958 earthquakes are those estimated by Mendoza
and Dewey [1984] with quality A or B. The black rectangle encloses the
high interplate coupling patch (C5) corresponding to the source of the
1958 earthquake.
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2209

Future studies should estimate the coupling strength in the entire 1979 source region and in our estimated
slip area of the 1906 earthquake near the trench. Such estimates are particularly important for improving our
understanding rupture processes along the Ecuador-Colombia subduction zone and for assessing seismic
and tsunami hazards in this coastal region.
References
Abe, K. (1979), Size of the great earthquakes of 1837–1974 inferred from tsunami data, J. Geophys. Res.,84, 1561–1568, doi:10.1029/
JB084iB04p01561.
Aki, K., and P. G. Richards (2002), Quantitative Seismology, Univ. Science Books, Sausalito, Calif.
Bouchon, B. (1981), A simple method to calculate Green’s functions for elastic layered media, Bull. Seismol. Soc. Am.,71, 959971.
Chlieh, M., et al. (2014), Distribution of discrete seismic asperities and aseismic slip along the Ecuadorian megathrust, Earth Planet. Sci. Lett.,
400, 292301.
Collot, J. Y., B. Marcaillou, F. Sage, F. Michaud, W. Agudelo, P. Charvis, D. Graindorge, M.-A. Gutscher, and G. Spence (2004), Are rupture zone
limits of great subduction earthquakes controlled by upper plate structures? Evidence from multi-channel seismic reflection data
acquired across the northern Ecuador–southwest Colombia margin, J. Geophys. Res.,109, B11103, doi:10.1029/2004JB003060.
Font, Y., M. Segovia, S. Vaca, and T. Theunissen (2013), Seismicity patterns along the Ecuadorian subduction zone: New constraints from
earthquake location in a 3-D a priori velocity model, Geophys. J. Int.,193, 263–286.
Fujii, Y., and K. Satake (2007), Tsunami source of the 2004 Sumatra-Andaman earthquake inferred from tide gauge and satellite data, Bull.
Seism. Soc. Am.,97, S192S207.
Fujii, Y., and K. Satake (2013), Slip distribution and seismic moment of the 2010 and 1960 Chilean earthquakes inferred from tsunami
waveforms and coastal geodetic data, Pure Appl. Geophys.,170, 1493–1509, doi:10.1007/s00024-012-0524-2.
Gailler, A., P. Charvis, and E. R. Flueh (2007), Segmentation of the Nazca and South American plates along the Ecuador subduction zone from
wide angle seismic profiles, Earth Planet. Sci. Lett.,260, 444–464.
Hashimoto, C., A. Noda, T. Sagiya, and M. Matsu’ura (2009), Interplate seismogenic zon es along the Kuril–Japan trench inferred from GPS data
inversion, Nat. Geosci.,2, 141144.
Hayes, G. P., D. J. Wald, and R. L. Johnson (2012), Slab1.0: A three-dimen sional model of global subduction zone geometries, J. Geophys. Res.,
117, B01302, doi:10.1029/2011JB008524.
Honda, K., T. Terada, Y. Yoshida, and D. Isitani (1908), Secondary undulations of oceanic tides, J. College Sci., Imper. Univ., Tokyo,26,1–113.
Kanamori, H., and K. C. McNally (1982), Variable rupture mode of the subduc tion zone along the Ecuador–Colombia coast, Bull. Seismol. Soc.
Am.,72, 1241–1253.
Kelleher, J. (1972), Rupture zones of large South American earthquakes and some predictions, J. Geophys. Res.,77, 2087–2103, doi:10.1029/
JB077i011p02087.
Kennett, B. L. N., and E. R. Engdahl (1991), Traveltimes for global earthquake location and phase identification, Geophys. J. Int.,105, 429465.
Kennett, B. L. N., E. R. Engdahl, and R. Buland (1995), Con straints on seismic velocities in the Earth from traveltimes, Geophys. J. Int.,122, 108–124.
Kikuchi, M., and H. Kanamori (1991), Inversion of complex body waves-III, Bull. Seism. Soc. Am.,81, 2335–2350.
Kikuchi, M., M. Nakamura, and K. Yoshikawa (2003), Source rupture processes of the 1944 Tonankai earthquake and the 1945 Mikawa
earthquake derived from low-gain seismograms, Earth Planets Space,55, 159–172.
Mendoza, C., and J. W. Dewey (1984), Seismicity associated with the great Colombia–Ecuador earthquakes of 1942, 1958 and 1979:
Implications for barrier models of earthquake rupture, Bull. Seismol. Soc. Am.,74, 577–593.
Moreno, M., M. Rosenau, and O. Oncken (2010), 2010 Maule earthquake slip correlates with pre-seismic locking of Andean subduction zone,
Nature,467, 198202.
Murotani, S., K. Satake, and Y. Fujii (2013), Scaling relations of seismic moment, rupture area, average slip, and asperity size for M~9
subduction-zone earthquakes, Geophys. Res. Lett.,40, 5070–5074, doi:10.1002/grl.50976.
Nakano, M., H. Kumagai, and H. Inoue (2008), Waveform inversion in the frequency domain for the simultaneous determination of
earthquake source mechanism and moment function, Geophys. J. Int.,173, 1000–1011.
Nocquet, J. M., et al. (2014) Motion of continental slivers and creeping subduction in the northern Andes, Nat. Geosci.,7, 287291,
doi:10.1038/ngeo2099.
Okada, Y. (1985), Surface deformation due to shear and tensile faults in a half-space, Bull. Seismol. Soc. Am.,75, 1435–1154.
Okal,E.A.(1992),UseofthemantlemagnitudeM
m
for the reassessmentof the moment of historical earthquakes, Pure Appl. Geophys.,139,1757.
Sakai, T., H. Kumagai, N. Pulido, J. Bonita, and M. Nakano (2016), Discriminating non-seismic long-period pulses and noise to improve
earthquake source inversion, Earth Planets Space,68, 50, doi:10.1186/s40623-016-0426-0.
Soloviev, S. L., and C. N. Go (1975), A catalogue of tsunamis on the eastern shore of the Pacific Ocean (1513–1968), translated by Canadian
Institute for Scientific and Technical Information, National Research Council Ottawa, Nauka House, Moscow.
Swenson, J. L., and S. L. Beck (1996), Historical 1942 Ecuador and 1942 Peru subduction earthquakes, and earthquake cycles along
Colombia–Ecuador and Peru subduction segments, Pure Appl. Geophys.,146,67–101.
Tanioka, Y., and K. Satake (1996), Tsunami generation by horizontal displacement of ocean bottom, Geophys. Res. Lett.,23, 861864,
doi:10.1029/96GL00736.
Tsuzuki, M., Koyama, J. and Yomogida, K. (2012), Was the 1906 great Ecuador-Colombia earthquake (M
w
8.8) a multiple rupture event of three
segments? paper presented at Japan Geoscience Union Meeting, Chiba, Japan.
Vallée, M., et al. (2013), Intense interface seismicity triggered by a shallow slow-slip event in the Central-Ecuador subduction zone, J. Geophys.
Res. Solid Earth,118,1–17, doi:10.1002/jgrb.50216.
Watada, S., S. Kusumoto, and K. Satake (2014), Traveltime delay and initial phase reversal of distant tsunamis coupled with the selfgravitating
elastic Earth, J. Geophys. Res. Solid Earth,119, 4287–4310, doi:10.1002/2013JB010841.
Wessel, P., W. H. F. Smith, R. Scharroo, J. Luis, and F. Wobbe (2013), Generic Mapping Tools: Improved version released, Eos Trans. AGU,94,
409410, doi:10.1002/2013EO450001.
White, S. M., R. Trenkamp, and J. N. Kellogg (2003), Recent crustal deformation and the earthquake cycle along the Ecuador–Colombia
subduction zone, Earth Planet. Sci. Lett.,216, 231–242.
Yoshimoto, M., S. Watada, Y. Fujii, and K. Satake (2016), Source estimate and tsunami forecast from far-field deepocean tsunami waveforms
—The 27 February 2010 M
w
8.8 Maule earthquake, Geophys. Res. Lett.,43, 659–665, doi:10.1002/2015GL067181.
Geophysical Research Letters 10.1002/2016GL071929
YOSHIMOTO ET AL. DEPTH-DEPENDENT RUPTURE MODE 2210
Acknowledgments
We obtained global seismic data from
the Data Management Center for the
Incorporated Research Institutes for
Seismology (IRIS-DMC). We obtained
tide gauge records of the 1906
earthquake at Honolulu and San
Francisco from the U.S. National Ocean
and Atmospheric Administration
(NOAA) and at Ayukawa from Honda
et al. [1908]. All figures were created
with the Generic Mapping Tools [Wessel
et al., 2013]. We thank Hugo Yepes and
Marta Calvache for their help in project
coordination. We are grateful to
Faustino Blanco for assisting with the
installation of the SWIFT system in
Colombia in its early stage and to Kenji
Satake for useful discussions about the
tsunamic source estimate of the 1906
earthquake. We thank Mohamed Chlieh
for sharing the plate coupling model
data. Comments from three anonymous
reviewers helped to improve the
manuscript. This work was supported by
the Science and Technology Research
Partnership for Sustainable
Development (SATREPS) project in
Colombia under the sponsorship of the
Japan Science and Technology Agency
(JST) and the Japan International
Cooperation Agency (JICA) and the JICA
technical cooperation project in
Ecuador.