Source Characteristics of the 22 January 2003 M

w
= 7.5 Tecoman, Mexico, Earthquake:
New Insights
LUIS QUINTANAR,
1
HE

CTOR E. RODRI

GUEZ-LOZOYA,
2
ROBERTO ORTEGA,
3
JUAN M. GO

MEZ-GONZA

LEZ,
4
TONATIUH DOMI

NGUEZ,
5
CLARA JAVIER,
6
LEONARDO ALCA

NTARA,
7
and CECILIO J. REBOLLAR
8
AbstractAftershock locations, source parameters and slip
distribution in the coupling zone between the overriding North
American and subducted Rivera and Cocos plates were calculated
for the 22 January 2003 Tecoman earthquake. Aftershock locations
lie north of the El Gordo Graben with a northwest-southeast trend
along the coast and superimposed on the rupture areas of the 1932
(M
w
= 8.2) and 1995 (M
w
= 8.0) earthquakes. The Tecoman
earthquake ruptured the northwest sector of the Colima gap, how-
ever, half of the gap remains unbroken. The aftershock area has a
rectangular shape of 42 2 by 56 2 km with a shallow dip of
roughly 12 of the Wadati-Benioff zone. Fault geometry calculated
with the Nabelek (1984) inversion procedure is: (strike, dip,
rake) = (277, 27, 78). From the teleseimic body wave spectra
and assuming a circular fault model, we estimated source duration
of 20 2 s, a stress drop of 5.4 2.5 MPa and a seismic moment
of 2.7 .7 9 10
20
Nm. The spatial slip distribution on the fault
plane was estimated using new additional near eld strong motion
data (54 km from the epicenter). We conrm their main conclu-
sions, however we found four zones of seismic moment release
clearly separated. One of them, not well dened before, is located
toward the coast down dip. This observation is the result of adding
newdata in the inversion. We calculated a maximumslip of 3.2 m, a
source duration of 30 s and a seismic moment of 1.88 9 10
20
Nm.
Key words: Rupture process, focal mechanism inversion,
waveform modeling.
1. Introduction
The subduction zone that encompasses the Mex-
ican states of Jalisco and Colima, (Fig. 1a) has
produced the largest earthquakes in Mexico. Offshore
of the state of Colima, the Rivera, Cocos and North
America plates interact (Fig. 1a). In this seismogenic
area the 1932, magnitude 8.2, Jalisco earthquake
occurred, the largest earthquake in Mexico in recent
documented seismological history (SINGH et al.,
1985; SUA

REZ et al., 1994). The 22 January 2003

(M
w
= 7.5) Tecoman earthquake occurred between
the rupture areas of the 9 October 1995 (M
w
= 8.0)
ColimaJalisco earthquake and the 30 January 1973
(M
w
= 7.6) Colima earthquake (see Fig. 1b) (REYES
et al., 1979; PACHECO et al., 1997). The Tecoman
earthquake was located by the University of Colima
at 18.625 N, 104.125 W, at a depth of 10 km and
with an origin time of 02 h 06 min and 35.0 s. This
earthquake badly damaged the state of Colima,
damaging and collapsing adobe houses, unreinforced
masonry structures and construction in poorly con-
solidated lls, and producing landslides. Liquefaction
was also observed in the Manzanillo port.
A preliminary study of the Tecoman earthquake
was carried out by SINGH et al., (2003). They
reported a preliminary aftershock area, gross source
characteristics, attenuation of strong motion data and
an isoseismal map. NU

EZ-CORNU et al., (2004)

reported the aftershock locations of the rst 72 h
after the main shock and found that the seismic
activity was mainly in the continental plate along
two NWSE trending zones. YAGI et al., (2004)
estimated the spatial and temporal slip over the fault
plane of the Tecoman earthquake, inverting regional
strong motion data as well as teleseismic body-
C. J. Rebollar: deceased September 2006.
1
Instituto de Geof sica, Universidad Nacional Autonoma de
Mexico, Mexico D.F., Mexico. E-mail: luisq@ollin.geosica.unam.mx
2
Facultad de Arquitectura, Facultad de Ingenier a Civil,
Universidad Autonoma de Sinaloa, Culiacan, Mexico.
3
Centro de Investigacion Cient ca y de Educacion Superior
de Ensenada, Unidad La Paz, La Paz, Mexico.
4
Centro de Geociencias, Universidad Nacional Autonoma de
Mexico, Juriquilla, Queretaro, Mexico.
5
Observatorio Vulcanologico de Colima, Universidad de
Colima, Colima, Mexico.
6
Comision Federal de Electricidad, Gerencia de Estudios de
Ingenier a Civil, Guadalajara, Mexico.
7
Instituto de Ingenier a, Universidad Nacional Autonoma de
Mexico, Mexico D.F., Mexico.
8
Centro de Investigacion Cient ca y de Educacion Superior
de Ensenada, Ensenada, Mexico.
Pure Appl. Geophys.
2010 Springer Basel AG
DOI 10.1007/s00024-010-0202-1
Pure and Applied Geophysics
waves. The closest near eld strong motion station
they used was located 138 km from the epicenter. In
the inversion process they used a slightly modied
fault plane geometry reported by the Harvard CMT
solution. They found three stages in the rupture
process: (1) the rupture nucleated near the hypo-
center, (2) the rst asperity, centering about 15 km
southwest from the epicenter, was broken and (3)
the rupture propagated to the northeast and the
second asperity was broken.
In the present study, we revise the aftershock area
employing newly calculated hypocenters using arri-
val times recorded at the Colima seismic network
(Red Sismica de Colima, RESCO) and nine portable
seismic stations deployed by CICESE (Centro de
Investigacion Cient ca y de Educacion Superior de
Figure 1
a Tectonic framework of the ColimaJalisco region. b Location of earthquakes [7 that have occurred in the ColimaJalisco region. Light
dashed lines show the rupture area of the 1932 (M
w
= 8.2) earthquake (SINGH et al., 1985). Heavy dashed line is the rupture area of the 1995
(M
w
= 8.0) earthquake (PACHECO et al., 1997). Also shown are the aftershock areas of the 1973 (M
w
= 7.3) and 1986 (1980) (M
w
= 7.0)
earthquakes. Continuous line is the aftershock area of the 2003 Tecoman earthquake. Available focal mechanisms are also shown. EPR east
Pacic rise
L. Quintanar et al. Pure Appl. Geophys.
Ensenada). We found that the aftershocks lay mainly
along the coupling zone of the plates and inside the
subducted slab. We also calculated the stress drops
and source duration of the main event with the
spectral analysis of body P-waves recorded by
broadband seismic stations located at teleseismic
distances, following BEZZEGHOUD et al., (1989),
assuming a circular fault model (BRUNE, 1970). In this
study of the Tecoman event, we used near eld strong
motion waveforms recorded at the Manzanillo power
plant located 54 km from the epicenter; data which
have not been previously used. We combined this
strong-motion data with the data used by YAGI et al.,
(2004) to obtain the slip on the fault plane and to
analyze how much the additional near eld data
modied the previously reported slip distribution on
the fault.
2. Tectonic and Historical Seismicity
West of the Mexican Volcanic Belt lies the
junction of the North American, Pacic and Cocos
plates. The Rivera and Cocos plates subduct at a
rate of 5 cm/year (KOSTOGLODOV and BANDY, 1995)
beneath North America, below the states of Colima
and Jalisco. In the North American plate a tectonic
block called the Jalisco Block (JB) is limited to the
west by the Rivera plate, to the south and southeast
by El Gordo-Colima graben and toward the east
and northeast by the Chapala and Tepic-Zacoalco
Grabens (Fig. 1a). Major seismic activity is due
mainly to the collision of the CocosRivera plates
with the North American plate and occurs in the
coupling zone. Depths of major underthrusting
earthquakes range from 10 to 32 km. According to
PARDO and SUA

REZ, (1995) seismic activity starts to

decrease toward Bahia Banderas, located northwest
of the JB, reaching depths on the order of 100 km
(see Fig. 1a). The Wadati-Benioff zone increases its
dip from *25 in the southeast to 46 toward
Bahia Banderas. Table 1 and Fig. 1b shows the
location, rupture areas, and seismic moments of
earthquakes of magnitude [7 that have occurred in
this region.
3. Data Recording Systems
The Volcanic Observatory of the University of
Colima maintains a short period and broadband tele-
metered network of seismic stations around the
Colima volcano and along the Colima coast. This
network, called Red S smica de Colima (RESCO),
consists of 11 short period seismometers and 2
broadband CMG-40T Guralp seismometers. Data are
continuously recorded and sent by radio links to the
University of Colima where they are stored and ana-
lyzed. RESCO stations, which record at 100 samples
per second, recorded the Tecoman event and all its
aftershocks. A week after the occurrence of the main
event, CICESE deployed a portable seismic network
of nine short period GBV seismic recorders with a
natural period of 4.5 Hz, as well as, a broadband
Guralp CMG-40T seismometer that recorded contin-
uously at 100 samples per second with a CMG-SAM2
acquisition module. All GBV seismic stations recor-
ded in a triggered mode at 100 samples per second.
Figure 2a shows the location of the seismic stations
and Fig. 2b the location of the strong motion stations.
4. Aftershock locations and dip of the Wadati-Benioff
zone
Aftershock locations were determined with the
HYPO71 program using a one-dimensional crustal
structure (shown in Table 2) routinely used by RE-
SCO. Due to the occurrence of the aftershocks toward
the ocean and the location of the seismic stations,
azimuthal coverage is poor. We located aftershocks
that included P and S wave arrival times at seismic
stations located along the coast in order to better
constrain focal depths. Figure 3 shows an example of
the traces of an aftershock recorded in 19 seismic
stations. Impulsive P wave arrivals as well as the
onsets of S waves can be seen. We started locating
aftershocks with initial xed depths in the range from
10 to 50 km with steps of 5 km. From the HYPO71
outputs we chose those locations with average root
mean square error of time residuals (rms) of
0.6 0.3 s, average standard error of the epicenter
Source Characteristics of Tecoman Earthquake
(ERH) of 3.3 1.8 km and average standard error of
the focal depth (ERZ) of 3.1 2.8. The errors are
plus or minus one standard deviation. Ninety after-
shocks fullled the above mentioned error criteria.
Local magnitudes of the located events range from 2
to 3.
Figure 4 shows the location of the aftershocks in a
plan view, from which it can be seen that the activity
lies north of El Gordo Graben and that the aftershock
area encompasses parts of the rupture areas of the
1932 and 1995 events. The area between the limits of
the rupture areas of the 1995 and the 1973 earth-
quakes is the Colima seismic gap. The northwest area
of this gap ruptured with the Tecoman earthquake.
Toward the southeast, roughly half of the gap remains
unbroken.
Figure 5 shows vertical cross-sections of after-
shocks projected perpendicular and parallel to the
middle America trench (MAT) as well as the pro-
jection of the P-axes of the Tecoman event and two of
the largest aftershocks of magnitudes 5.7 and 5.3.
Also shown is the projection of the 1995 M
w
= 8.0
ColimaJalisco event. The hypocenter (where the
rupture started) of the Tecoman event lies within the
coupling zone between the subducted CocosRivera
plates and the overriding North American plate. The
boundary between the subducted slab and the conti-
nental crust was taken from cross section K of
PARDO and SUA

REZ, (1995). The source mechanisms

of the main event and two of the largest aftershocks
were all thrust events. The projections of the P-axes
of the main event and the largest aftershocks show
that these compressional axes are oblique to the
interface of the overriding and subducted slab. One of
the aftershocks (focal depth of 5 km) was an inter-
plate event, while the second large aftershock (focal
depth of 14 km) was an intraplate event. Fault plane
solutions of the largest aftershocks were taken from
SINGH et al., (2003). Using the obtained aftershocks
depth, we inferred that the Wadati-Benioff zone in
the coupling area has a dip of 12 up to a distance of
85 km from the trench (see Fig. 5a); beyond that
point the Wadati-Benioff increases to a dip of 50
according to PARDO and SUA

REZ, (1995). Most of the

aftershock activity of the 2003 Tecoman sequence
occurred within the subducted plate and along the
overriding North American plate. The thick line on
Fig. 5a depicts our interpretation of the coupling
zone. On the other hand, Fig. 5b shows a cross-sec-
tion parallel to the MAT, from this gure we roughly
calculated a rupture area of 44 by 58 km. Aftershock
depths range from 5 to 38 km.
5. Focal Mechanism Determination
To obtain the fault geometry we carried out a
waveform inversion with the point source approxi-
mation using P and SH recorded at teleseismic
distances following Nabelek (1984). We used data
from stations at epicentral distances between
30 \D
P
\91 and 35 \D
SH
\83 for P and SH
waves, respectively. Various corrections were inclu-
ded for a direct comparison between synthetic and
observed seismograms. Such corrections take care of
the geometrical spreading, the instrumental response,
Table 1
Location and seismic moments of earthquakes of magnitude [7 that have occurred in the study region
Date H:min. Lat. Lon. Seismic moment (Nm) Magnitude (Ms
c
, M
w
a
) Depth (Km)
19000120 06:33 20.00 -105.00 7.4
c

19320603 10:36 19.80 -104.00 9.10 9 10

20
8.2
c
, 8.0
a
16.0
19320618 10:12 18.95 -104.42 7.3 9 10
20
7.8
c
16.0
19410415 19:09 18.85 -102.94 7.7
19730130 21:01:00 18.39 -103.21 3.0 9 10
20
7.3
bc
32.0
19950910 15:35:53.0 19.05 -104.20 9.1 9 10
20
8.0 17.0
20030122 02:06:33.8 18.625 -104.125 1.60 9 10
20
7.4
a
10.0
a
RESCO
b
Reyes et al. (1979)
c
Pacheco and Sykes (1992)
L. Quintanar et al. Pure Appl. Geophys.
the amplitude and phase distortions which are due
to a layered crust. The anelastic attenuation of the
Earth (FUTTERMAN, 1962) is responsible for these
distortions. To correct for it, we used a frequency
independent Q average seismic quality factor along
the ray, which is assumed a constant value of t* = T/Q,
where T is the ray travel time. Here we used t* = 1.0
and 4.0 s for P and SH waves, respectively.
The effects of the structure below the receiver
were neglected and we consider the response to be
Figure 2
a Location of RESCO (full triangles) seismic stations and CICESE (crosses) portable stations. b Map showing the strong motion instruments
from UNAM (full squares). Empty square shows the location of the Manzanillo strong-motion recorder located in the Manzanillo power plant
Table 2
Velocity model used for aftershocks location
P-wave velocity (km/s) Layer depth (km)
1.7 0.0
2.7 2.0
3.6 3.0
4.6 4.0
5.7 6.0
6.0 12.0
7.4 18.0
7.8 35.0
Source Characteristics of Tecoman Earthquake
controlled mostly by the source radiation pattern at
the near source structure. Greens functions around
the source were computed and a stratied velocity
model was assumed (Table 3) based on the crustal
structure of REYES et al., (1979). A simple half-space
model at the receiver was used (V
p
= 6.0 km/s,
V
s
= 3.46 km/s, q = 2.8 g/cm
3
and a Poisson ratio
of 0.25). Seismograms were band-pass ltered using
a third-order Butterworth lter from 0.01 to 0.4 Hz
and corrected for instrument responses. Sixty seconds
of vertical P and transverse SH wave displacements
were obtained through integration of the velocity
records. Good station coverage made it possible to
obtain the fault geometry. The event was modeled
using 7 overlapping triangles with 2.5 s of rise time.
We obtained the following mechanism: (strike, dip,
rake) = (277, 27, 78), and a scalar seismic
moment of 1.3 9 10
20
Nm and a apparent source
time function of 20 s. By searching for the best
centroid depth, the residual error minimized at
25 km, which is similar to that reported by YAGI
et al., (2004). Waveform tting was good even for
SH waves, which are usually difcult to t. Figure 6
shows a comparison between observed and synthetic
waveforms and the apparent source time function. P-
wave tting around the fault plane is reasonably
good. Southward stations, in the dilatation quadrant
close to the sub-vertical nodal plane (PTCN and
RPN), constrained the strike of the south dipping
plane. This is important because most of the stations
Figure 3
Example of an aftershock recorded by the 19 seismic stations (only the vertical component is shown). Impulsive P-wave arrivals and clear
onsets of some S waves can be seen; traces are normalized to one. Names of the seismic stations are indicated
L. Quintanar et al. Pure Appl. Geophys.
were distributed into the compression quadrant. A
similar good quality tting was found for SH-waves;
stations PTCN, DSB, HRV, and MTE xed one of the
nodal planes. The NA

BE

LEK, (1984) point source

waveform tting was good to constrain the fault
geometry, however, it clearly underestimated the
seismic moment and the apparent source time func-
tion (also shown in the middle). Finite two-dimension
source inversion (described later) showed that some
patches of slip are detected by near eld stations
which raised the scalar moment, and that displace-
ment could not be detected by the point source
approximation at teleseismic distances, especially
those displacements near the edge of the fault plane.
6. Static Stress Drop Estimate
We calculated the P-wave spectra of the Tecoman
earthquake recorded at broadband stations at telese-
ismic distances. We used IRIS and GEOSCOPE
seismic stations in order to estimate source parame-
ters. P-wave windows of 70 s, which included 10 s of
noise prior to the P-wave arrival and 60 s of the P
wave trains, were used in the spectral analysis. No
lter was applied to the P wave trains. After calcu-
lating the fast Fourier transform, we estimated the
seismic moment rate at each seismic station by:
o
ot
M f
4pqa
3
a
g D; h Ci
Q f W f
I f S
rad
f

1
where q is the density at the source, a is the P wave
velocity at the source, a is the radius of the earth,
C(i) is the surface amplication at the ith seismic
station, Q(f) is the seismic attenuation, g(D, h) is the
geometrical dispersion, W(f) is the spectral displace-
ment, I(f) is the instrument response and S
rad
(f) is the
radiation pattern as a function of frequency of the P,
pP and sP waves.
Data was corrected for instrument response,
attenuation, and radiation pattern. Following BEZ-
ZEGHOUD et al., (1989), we substituted the effective
radiation pattern S
rad
(f) with the approximation:
S
rad
0 R
P
R
pP
R
sP
2
where R
P
, R
pP
, R
sP
are the P, pP, and sP radiation
pattern, respectively. The fault geometry used to
calculate the radiation pattern was strike 277, dip
27 and rake 78. According to the hypocenter
(25 km), the pP and sP wave trains arrived 7 and 10 s
after the direct P wave. The attenuation operator t*
was assumed to be 1.0. The static seismic moment
was calculated from the following expression:
M
0

o
ot
M0
4pqa
3
a
g D; h Ci
Q0W0
I0S
rad
0

3
Figure 7 shows the plot of the logarithmic mean
of corrected moment rate spectra determined from the
broadband displacement records of 15 P-waveforms
recorded at teleseismic stations. An average seismic
moment of 2.7 0.7 9 10
20
Nm (2.7 0.7 9 10
27
dynes cm) was determined. From the spectra we
obtained a corner frequency of f = 0.05 0.005 Hz
and a slope at higher frequencies of -2.4. The rup-
ture time needed for the rupture to propagate along
the fault is approximately:
T
R

L
v
R
4
Where T
R
is the rupture time, which is the inverse
on the corner frequency, L is the length of the fault
Figure 4
Map showing our best aftershock locations (empty circles). Large
full square shows the epicenter location of the Tecoman earth-
quake. Its fault plane solution is also shown. Large empty square
shows the epicenter location of the ColimaJalisco earthquake of
1995 (M
w
= 8.0). Proles AA
0
and BB
0
are cross sections onto
which aftershocks were projected shown in Fig. 5. Triangles show
the locations of seismic stations
Source Characteristics of Tecoman Earthquake
and v
R
is the rupture velocity. Therefore, the rupture
time is 20 2 s and the rupture length is 66 5 km,
assuming a rupture velocity of 0.9V
s
(3.3 km/s). This
calculated rupture length lies within the range of the
aftershock length calculated with the aftershock
activity (56 2 km). The stress drop on a circular
fault (BRUNE, 1970) with radius R is given by
Dr
7
16
M
0
R
3
5
From the aftershock area we calculated a fault
length of 56 2 km so the fault radius R is
28 2 km. Therefore, the stress drop of the
Tecoman earthquake is 5.4 2.5 MPa. This value
contrasts with the higher value (*10 MPa) obtained
by YAGI et al., (2004) using regional and local
records.
7. Slip Distribution Inversion
BERESNEV, (2003) argued that incorrect assump-
tions of crustal structure, rupture velocity, fault
geometry and array geometry will generate geologic
artifacts in the inversion of seismic data for slip
Figure 5
a Cross section of seismicity along prole AA
0
. b cross section of seismicity along prole BB
0
. Arrows show the projection of the P axes for
the Tecoman earthquakes and two of its largest aftershocks. Subducting plate geometry was taken from PARDO and SUA

REZ, (1995). See Fig. 4

for prole locations
Table 3
Velocity model used for focal mechanism determination with
teleseismic records
Thickness (km) V
p
(km/s) V
s
(km/s) q (g/cm
3
)
6.0 5.80 3.35 2.68
19.0 6.40 3.69 2.78
10.0 7.00 4.04 2.85
8.00 4.62 3.00
L. Quintanar et al. Pure Appl. Geophys.
distribution of nite faults. On the other hand, IDE
et al., (2005) suggested that the quality and quantity
of the data is critical to constrain the source-imaging
problem. So, in line with those opinions, we decided
to do the source imaging of the Tecoman event with
better-constrained fault geometry and the inclusion of
near eld strong motion data not used previously in
the study of the Tecoman event. Figure 8 shows the
acceleration records of the free-eld strong motion
station located at the Manzanillo power plant (MNZ).
We also used data recorded at IRIS and GEOSCOPE
broadband stations, strong motion records from the
Servicio Sismologico Nacional (SSN) and Instituto
de Ingenieria, UNAM (Fig. 2). To obtain the slip
over the rupture area of the Tecoman earthquake, we
used the method described by YAGI et al., (1999,
2004) following the formulation of HARTZELL and
HEATON, (1983), YOSHIDA (1989, 1992). The rupture
process is simulated as a spatio-temporal slip distri-
bution over the fault plane. The fault plane is divided
Figure 6
Point source inversion from body wave broadband data. a Observed P wave seismograms (black lines) and synthetics (red lines). The source
time function was about 20 s and the corresponding seismic moment is 1.3 9 10
20
Nm. b Observed SH wave seismograms (solid lines) and
synthetics (dashed lines). The solution corresponds to an underthrust event with (strike, dip, rake) = (277, 27, 78). The axes indicate the
amplitude of the waves in microns
Source Characteristics of Tecoman Earthquake
into M 9 N sub-faults with length xx and width yy.
The slip-rate function at each sub-fault is described
with B splines that are a series of L triangle functions
with rise time s. Slip vectors at each sub-fault are
taken as vectors with components given by k
0
, the
assumed slip angle. The synthetic waveform at the jth
station is given by
W
obs
j
t
i

X
m
X
n
X
l
X
k
X
mnlk
g
mnkj
t
i
l 1 s T
mn
e;
where X
mnlk
is the lth component of slip at mnth
subfault at kth time step. g
mnkj
is the Greens function
from a point source at the mn-subfault, T
mn
is the start
time of the slip of the basis functions at each sub-fault
and e is assumed to be a Gaussian error with variance
r. Greens functions for teleseismic body waves are
calculated using the KIKUCHI and KANAMORI, (1991)
method. Greens functions for near eld ground
motion are calculated by the discrete wave number
method developed by KOHKETSU, (1985). Table 4
shows the one-dimensional elastic models used in the
calculation of the Greens functions. This model is
the same used by YAGI et al. (2004) for teleseismic
stations and has been modied lightly for strong
motion stations to take into account the inclusion of
the MNZ accelerograms.
In the inversion process we used the fault geom-
etry previously calculated with the NA

BE

LEK, (1984)
method given by (strike, dip, slip) = (277, 27, 78).
We used a larger fault area in order to follow the
same procedure of YAGI et al., (2004), even though
the observed aftershock area is smaller. We consid-
ered a rupture area with a width of 70 km, a length of
100 km and divided the fault plane into 14 9 20 sub-
faults with length of 5 km and width of 5 km. Twenty
ve kilometers depth of the initial front rupture was
assumed. The slip-rate function on each sub-fault was
expanded into a series of twelve triangles with a rise
time of 2.0 s and the slip angle kept unchanged
during the rupture. We used a rupture wave velocity
of 3.3 km/s.
It is instructive to investigate how well the MNZ
accelerogram ts with our estimated fault geometry
(strike, dip, slip) = (277, 27, 78) and YAGI et al.
(2004) fault geometry (strike, dip, slip) = (300, 20,
93). Figure 9 shows the comparison of the observed
and synthetic accelerograms of MNZ with both fault
Figure 7
Average displacement spectra of teleseismic P waveforms of the Tecoman earthquake. f is the corner frequency and dashed lines indicate the
trend at lower and higher frequencies. The slope at higher frequencies is -2.4
L. Quintanar et al. Pure Appl. Geophys.
geometries. Both geometries reproduce the main
amplitudes, but our fault geometry t better the later
parts of the accelerograms of the vertical and north
south components; however both geometries failed to
adequately t the eastwest component, even though
our fault geometry better t the initial onset of the
rst arrivals.
We performed the inversion with the near eld
strong motion data and the far eld P-waveforms.
Comparisons of the observed and synthetic wave-
forms, shown in Fig. 10, indicate that the ts are
good. The seismic moment was 1.88 9 10
20
Nm
equivalent to a moment magnitude of 7.5. The source
time duration lasted nearly 30 s, but the majority of
the energy was released during the rst 20 s, in
agreement with the source time duration calculated
with the teleseimic spectra and NA

BE

LEK (1984) point

source inversion. Figure 11 show a horizontal pro-
jection of the slip distribution on the fault plane,
which consisted mainly of the same seismic moment
release found by YAGI et al., (2004), however, it
shows two additional zones of seismic moment
release up-dip and down-dip which are barely shown
in YAGI et al., (2004) inversion. The maximum fault
slip was 3.2 m.
8. Summary
We used data recorded by a permanent network
of short period and broadband seismic stations
(RESCO) and nine portable seismic stations in order
determine the aftershock area of the Tecoman earth-
quake. Figure 4 shows our aftershock locations. A
depth of 10 km for the hypocenter of the Tecoman
earthquake was calculated with P-wave arrivals of
local stations; we think that the rupture could have
started at this depth. The depths of the main event and
the aftershocks dened a shallow dip of the Wadati-
Benioff zone of roughly 12 (Fig. 5). Beyond the
aftershock area, the Wadati-Benioff dip increases to
50 according to PARDO and SUA

REZ, (1995), however

we did not located any aftershock activity at the bend
of the subducted slab. We inferred an approximate
aftershock area with a rectangular shape of 42 2 by
56 2 km. Aftershocks were located toward the
north of the location of the epicenter of the Tecoman
Figure 8
Acelerograms recorded in the Manzanillo (MNZ) power plant
located 54 km from the epicenter
Table 4
Velocity model used for rupture process inversion with local,
regional and teleseismic records
a (km/s) b (km/s) q (10
3
kg/m
3
) Qa Qb Thickness
(km)
For teleseismic body wave
1.5 0.0 1.0 1.0
5.54 3.20 2.50 3.0
5.69 3.29 2.70 5.0
6.27 3.62 2.80 9.70
6.71 3.87 2.90 17.30
8.10 4.68 3.30
For strong ground motion
5.69 3.37 2.70 500 250 9.0
6.27 3.54 2.80 600 300 9.7
6.71 3.82 2.90 800 400 17.3
8.00 4.53 3.25 1,200 600
Source Characteristics of Tecoman Earthquake
event, and they enclose the southeast aftershock area
of the 9 October 1995 (M
w
= 8.0) ColimaJalisco
earthquake. The Tecoman earthquake ruptured
approximately half of the Colima gap. Most after-
shocks occurred inside the subducted slab, and our
aftershock locations (Fig. 5) have features similar to
those reported by NU

EZ-CORNU et al., (2004) and

YAGI et al., (2004). For example, they report an
elongated trend of aftershocks from southeast to
northwest and three clusters of activity; however,
their locations have a trend toward the southwest. Our
aftershock locations are spread in an extended area
close to the seashore with some of the features
reported by NU

EZ-CORNU et al., (2004) and YAGI

et al., (2004).
Source parameters of the Tecoman earthquake
were calculated from the spectra of P-waveforms
recorded at teleseismic distances. A corner frequency
was observed at f = 0.05 005 Hz, a stress drop of
5.4 2.5 MPa (54 25 bars) and a seismic moment
of 2.7 0.70 9 10
20
Nm (2.7 0.7 9 10
27
dynes
cm) were calculated assuming that BRUNES (1970)
circular fault model holds. We note the lower stress
drop value calculated with teleseismic data compared
with that obtained using local and regional records
(YAGI et al., 2004, SINGH et al., 2003). This result
would imply that Tecoman earthquake was decient
in high-frequency radiation. This, however, appears
not to be the case from the reported severe damage to
buildings in the localities situated at distances where
along-trench rupturing subduction earthquakes of
the same magnitude usually do not produce any
signicant damage (ZOBIN and PIZANO-SILVA, 2007).
We think, therefore, that a low rupture velocity is
more likely the cause of the low stress drop value
obtained from spectral analysis.
Besides the waveforms used by YAGI et al., (2004)
we included, in the inversion of near eld body
waves, the waveforms recorded at the Manzanillo
power plant (MNZ), a strong motion station located
54.0 km from the epicenter. Those body waveform
were not used by YAGI et al., (2004). In the inversion
process we assumed a rectangular fault area of 70 by
100 km and a depth of the initial break of 25 km.
Figure 11 shows a horizontal projection of the fault
rupture from our inversion. The slip distribution
shows the same asperities reported by YAGI et al.,
(2004), however, two additional zones of seismic
moment release, not so energetic in their inversion,
appeared clearly dened in our analysis as a signi-
cant zone of seismic moment release up-dip and
down-dip. The total source duration was 30 s and a
seismic moment of 1.88 9 10
20
Nm (M
w
= 7.5). A
signicant seismic moment release was observed at
the end of the rupture process; this seismic moment
released was located down-dip below the Manzanillo
seismic station and up-dip. The source time duration
of 20 s calculated with the spectra of teleseismic
body-waves and with the inversion of P and SH
waves using the NA

BE

LEK, (1984) method agree with

the main source energy release. On the other hand,
source time duration obtained with the inversion of
teleseismic and near-source strong motion stations is
30 s, however the maximum seismic moment release
Figure 9
Comparison of modeling for the strong-motion record of MNZ station located 54 km from the epicenter. a Modeling with the focal
mechanism solution obtained in this work using Nabelek technique. b Results obtained using YAGI et al., (2004) geometry
L. Quintanar et al. Pure Appl. Geophys.
lasted about 20 s (see Fig. 11b). We think that the
large source duration obtained with the two dimen-
sional method of inversion is due to the effect of the
contribution of energy recorded on the strong-motion
(MNZ) stations which was not observed at teleseis-
mic stations.
It is worth mentioning that the uncertainties sug-
gested by BERESNEV, (2003) about the slip inversions
using body waves did not have much effect in our
study. We did not nd great disagreement in the
inversion process for the spatial slip distribution. On
the other hand, we found an improvement in the
Figure 10
Comparison of the observed waveforms (black line) with the calculated waveforms (red line) obtained from the inversion. Seven upper
records correspond to strong motion data of Mexican stations. Six lower seismograms are teleseismic records. Numbers below the station
name are the maximum amplitude in centimeters and microns of strong-motion and teleseismic records, respectively
Source Characteristics of Tecoman Earthquake
spatial slip distribution by adding the new near source
body waveforms into the inversion, therefore, the
quality and quantity of waveforms in the near and far
eld will greatly improve our knowledge of the
source process as suggested by IDE et al., (2005).
Acknowledgments
We would like to thank Arturo Perez-Vertti for the
eldwork and Antonio Mendoza for helping to handle
the digital data. Thanks to the following networks for
Figure 11
a Fault plane solution used for the slip distribution inversion on the fault plane. b Source time function obtained in this inversion. c Horizontal
projection of the fault showing the slip distribution. Triangle shows the location of the Manzanillo power plant, where it is located the MNZ
strong motion recorder, Colima and Tecoman cities
L. Quintanar et al. Pure Appl. Geophys.
providing seismic data through the Incorporated
Research Institutions for Seismology (IRIS) Data
Management Center: FDSN, GEOFON, Geoscope,
IDA, GTSN, UI, USGS. We acknowledge El Consejo
Nacional de Ciencia y Tecnologia (CONACYT) for
sponsoring part of this project (Proyecto 37038-T).
One of us (HERL) was supported by a scholarship
from PROMEP.
REFERENCES
BERESNEV, I. A. (2003), Uncertainties in Finite-Fault Slip Inver-
sion: To What Extent to Believe?, Bull. Seismol. Soc. Am. 93,
24452458.
BEZZEGHOUD, M., DESCHAMPS, A. and MADARIAGA, R., Broadband
P-wave Signals and Spectra from Digital stations. In Digital
Seismology and Fine Modeling of the Lithosphere (eds. Cassinis,
R., Nolet, G., and Panza, G.F.) (Plenum Press, New York,
351374, 1989).
BRUNE, J.N. (1970), Tectonic stress and the spectra of seismic shear
waves from earthquakes, J. Geophys Res. 75, 49975009.
FUTTERMAN, W.I. (1962). Dispersive body waves, J. Geophys. Res.
67, 52795291.
SUA

REZ. G., GARCIAACOSTA, V. and GAULON, R. (1994), Active

crustal deformation in the Block, Mexico: evidence for great
historical earthquake in the 16th century, Tectonophysics 234,
117127.
HARTZELL, S. H. and HEATON, T.H. (1983), Inversion of strong
ground motion and teleseismic waveform data for the fault
rupture history of the 1979 Imperial Valley, California earth-
quake, Bull. Seismol. Soc. Am. 73, 15531583.
IDE, S., BEROZA, G.C. and MCGUIRE, J.J., Imaging earthquake
source complexity, Geophysical Monograph 157, Seismic earth:
Array analysis of broadband seismograms (eds. Levender, A.
and Nolet, G. (American Geophysical Union, Washington, D. C.,
pp, 2005).
KIKUCHI, M., and KANAMORI, H. (1991), Inversion of Complex Body
Waves-III, Bull. Seism. Soc. Am. 81, 23352350.
KOHKETSU, K. (1985), The extended reectivity method for synthetic
near-eld seismograms, J. Phys. Earth 33, 121131.
KOSTOGLODOV, V., and BANDY, W. (1995), Seismotectonic con-
straints on the rate between the Rivera and North American
plates, J. Geophys. Res. 100, 977990.
NA

BE

LEK, J. (1984), Determination of earthquake source parame-

ters from inversion of body waves, PhD. Thesis (Mass. Inst. of
Technol., Cambridge), 346 pp.
NU

EZ-CORNU, F. J., REYES-DA

VILA, G.A., RUTZ LO

PEZ, M., TREJO

GOMEZ, R., CAMARENA-GARCI

A, M.A. and RAMI

REZ-VAZQUEZ, C.A.
(2004), The 2003 Armeria, Mexico Earthquake (M
w
= 7.4):
Mainshock and early aftershocks, Seismological Research Let-
ters 75, pp. 506605.
PACHECO, J. F., and SYKES, L. (1992), Seismic moment catalog of
large shallow earthquakes, 1900 to 1989, Bull. Seismol. Soc.
Am. 82, 13061349.
PACHECO, J., SINGH, S. K., DOMINGUEZ, J., HURTADO, A., QUINTANAR,
L., JIME

NEZ, Z., YAMAMOTO, J., GUTIE

RREZ, C., SANTOYO, M.,

BANDY, W., GUZMAN, M., KOSTOGLODOV, V., REYES, G. and
RAMI

REZ, C. (1997), The October 9, 1995 ColimaJalisco,

Mexico earthquake (Mw 8): an aftershock study and comparison
of this earthquake with those of 1932, Geophys. Res. Lett. 24,
22232226.
PARDO, M., and SUA

REZ, G. (1995), Shape of the subducted Rivera

and Cocos plates in Southern Mexico: Seismic and tectonic
implications, J. Geophys. Res. 100, 1235712373.
REYES, A., BRUNE, J. N. and LOMNITZ, C. (1979), Source mechanism
and aftershock study of the Colima, Mexico earthquake of Jan-
uary 30, 1973, Bull. Seismol. Soc. Am. 69, 18191840.
SINGH, S.K., PONCE, L. and NISHENKO, S. P. (1985), The great
Jalisco, Mexico, earthquakes of 1932: subduction of the Rivera
plate, Bull. Seism. Soc. Am. 75, 13011313.
SINGH, S. K., PACHECO, J. F., ALCA

NTARA, L., REYES, G., ORDAZ, M.,

IGLESIAS, A., ALCOCER, S. M., GUTIERREZ, C., VALDE

S, C., KOS-
TOGLODOV, V., REYES, C., MIKUMO, T., QUAAS, R., ANDERSON, J.G.
(2003), Contributions. A Preliminary Report on the Tecoman,
Mexico Earthquake of 22 January 2003 (Mw 7.4) and Its Effects.
Seism. Res. Lett. 74, 279289.
YAGI, Y., KIKUCHI, M., YOSHIDA, S. (1999), Comparison of the
coseismic rupture with the aftershock distribution in the Hyuga-
nada earthquakes of 1996, Geophys. Res. Lett. 26, 31613164.
YAGI, Y., MIKUMO, T., PACHECO, J., and REYES, G. (2004), Source
rupture process of Tecoman, Colima, Mexico earthquake of
January 22, 2003, determined by joint inversion of teleseismic
body-wave and near-source data, Bull. Seismol. Soc. Am. 94,
17951807.
YOSHIDA, S. (1989), Waveform inversion using ABIC for the rupture
process of the 1983 Hindu-Kush earthquake, Phys. Earth Planet.
Inter. 56, 389405.
YOSHIDA, S. (1992), Waveform inversion for rupture process using
a non-at seaoor model: application to 1986 Andreanof Islands
and 1985 Chile earthquakes, Tectonophysics 211, 4559.
ZOBIN, V.M. and PIZANO-SILVA, J.A. (2007), Macroseismic Study of
the M
w
7.5 21 January 2003 Colima, Mexico, Across-Trench
Earthquake, Bull. Seismol. Soc. Am. 97, 12211832.
(Received February 11, 2010, revised September 22, 2010, accepted September 23, 2010)
Source Characteristics of Tecoman Earthquake