Nat Hazards (2013) 69:1261–1279

DOI 10.1007/s11069-013-0629-7

ORIGINAL PAPER

Numerical simulation of dynamic mechanisms of the

2008 Wenchuan Ms8.0 earthquake: implications
for earthquake prediction

Shoubiao Zhu

Received: 5 August 2012 / Accepted: 4 March 2013 / Published online: 13 March 2013
Ó Springer Science+Business Media Dordrecht 2013

Abstract The sudden and unexpected Wenchuan earthquake (Ms = 8.0) occurred on the
Longmen Shan Fault, causing a large number of casualties and huge property loss. Almost
no definite precursors were reported prior to this event by Chinese scientists, who made a
first successful prediction of the 1975 Haicheng earthquake (M = 7.3) in China. Does the
unsuccessful prediction of the Wenchuan earthquake mean earthquake prediction is
inherently impossible? In order to answer this question, the paper simulated inter- and
co-seismic deformation, and recurrence of strong earthquakes associated with the Long-
men Shan listric thrust fault by means of viscoelastic finite element method. The modeling
results show that the computed interseismic strain accumulation in the lower crust beneath
the Eastern Tibet is much faster than that in the other regions. In particular, the elastic
strain energy density rate accumulates very rapid in and around the Longmen Shan fault in
the depth above *25 km that may explain why the great Wenchuan earthquake occurs in
the region of such a slow surface deformation rate. The modeled coseismic displacements
around the fault are consistent with surface rupture, aftershock distribution, and GPS
measurement. Also, the model displays the slip history on the Longmen Shan fault,
implying that the average earthquake recurrence interval on the Longmen Shan fault is
very long, 3,300 years, which is in good agreement with the observed by paleoseismo-
logical investigations and estimates by other methods. Moreover, the model results indicate
that the future earthquake could be evaluated based on numerical computation, rather than
on precursors or on statistics. Numerical earthquake prediction (NEP) seems to be a
promising avenue to a successful prediction, which will play an important part in natural
hazard mitigation. NEP is difficult but possible, which needs well supporting.

S. Zhu (&)
Institute of Crustal Dynamics, China Earthquake Administration, No. 1, Road Anningzhuan,
Haidian District 100085, Beijing, China
e-mail: [email protected]

S. Zhu
Key Lab of Computational Geodynamics, Chinese Academy of Sciences, No.19A,
Road Yuquan, Beijing 100049, China

123
1262 Nat Hazards (2013) 69:1261–1279

Keywords Dynamical mechanisms
Longmen Shan fault
Finite element method
Inter-
and coseismic deformation
Numerical earthquake prediction
Wenchuan earthquake

1 Introduction

On May 12, 2008, a devastating earthquake struck densely populated Wenchuan county,
Sichuan Province, China (31.0°N, 103.4°E) (www.csi.ac.cn/sichuan). More than 80,000
people were killed, over 370,000 people were injured, and economic losses worth more
than 800 billion RMB (nearly US$100 billion) have been estimated (www.csi.ac.cn/
sichuan). The Wenchuan earthquake is the most disastrous event in China since the 1976
Tangshan earthquake, which killed over 240,000 people.
The sudden and unexpected Wenchuan earthquake occurred on the Longmen Shan
thrust belt, the transition zone between the Tibetan Plateau and the Sichuan basin.
Although the steep western margin of the Sichuan basin is known to be seismically active,
few, if any, earth scientists anticipated an event of this magnitude there. At least three main
observations are responsible for this biased viewpoint. First of all, on the Longmen Shan
fault belt, there have been no earthquakes with magnitude greater than 7 occurred in the
long history of human civilization (Wen et al. 2008, 2009). Second, Global Positioning
System (GPS) measurements (King et al. 1997; Chen et al. 2000; Shen et al. 2005; Zhang
et al. 2004; Gan et al. 2007) and active faulting studies (Burchfiel et al. 1995, 2008;
Densmore et al. 2007; Zhou et al. 2007) reveal very slow slip rates (\2–3 mm/yr) across
the Longmen Shan fault zone, indicative of relatively modest strain accumulation and
therefore a slowly accumulating seismic hazard. And last, there were no findings of any
anomaly for earthquake precursors before the Wenchuan earthquake.
Obviously, China neither made any short-term prediction nor made any intermediate-
term forecasting to this strong event. The main reason may be that there were almost no
precursors such as foreshocks, rapid changes of surface strain, changes of water levels in
wells, variations in electromagnetic properties of the earth, and strange behaviors of certain
animals and human beings. In fact, it is not surprising that large earthquakes had few
precursors. As early as in 1997, Kagan pointed out that all reported cases of precursors to
large earthquakes can be explained by random noise or by chance coincidence, no
statistically rigorous validation of proposed precursory phenomena was available despite
considerable effort in several countries (Kagan 1997). In other words, no successful
earthquake prediction is made only by precursors to date. Recently, some strong earth-
quakes occurred in the world, for example, the April 14, 2010, Yushu, China earthquake
(Ms = 7.1), February 27, 2010, Chile earthquake (Mw = 8.8), the January 12, 2010, Haidi
earthquake (Mw = 7.0), and the March 11, 2012, Japanese Tohoku earthquake
(Mw = 9.0). No alarm warning was given out for any of these strong events, which has
negative impacts both on scientific community and on society. Then, does it corroborate
Geller’s viewpoint (Geller 1997; Geller et al. 1997): ‘‘Earthquake cannot be predicted’’?
In order to answer the above question, it is necessary to understand why the slowly
deforming Longmen Shan fault zone produced such a strong earthquake. And what are the
seismic cycles on the Longmen Shan fault belt? In this paper, viscoelastic finite element
method is used to simulate the dynamic processes of the Wenchuan earthquake and to
explore an alternative approach to earthquake prediction.

123
Nat Hazards (2013) 69:1261–1279 1263

2 Seismogenic structure of the Wenchuan earthquake

The Wenchuan earthquake occurred on the Longmen Shan thrust fault, which is located at the
conjunction between the eastern edge of the Tibetan Plateau and the western margin of
the Sichuan basin. On the eastern side of the Tibet, the Longmen Shan rise 6,000 m above the
Sichuan basin, exhibiting greater relief than anywhere else on the plateau (Hubbard and Shaw
2009). The active Longmen Shan fault zone marks a predominantly convergent boundary
with a right-lateral strike slip component. This fault system was reactivated during late
Cenozoic time along a Mesozoic orogenic belt (Burchfiel et al. 1995, 2008; Kirby et al. 2008).
To the west, the Eastern Tibet (Songpan–Ganzi Terrain in geological terminology) actively
deforms by both right-lateral shearing and convergence perpendicular to the Longmen Shan
fault (King et al. 1997; Chen et al. 2000; Shen et al. 2005; Zhang et al. 2004; Gan et al. 2007).
In contrast, tectonic and seismic activity in the Sichuan basin, east of the Longmen Shan, has
been mild during late Cenozoic time (shown in Fig. 1).

Fig. 1 Active tectonic map of western Sichuan region. Yellow star denotes location of the 2008 Wenchuan
earthquake. Blue arrows are GPS velocity with respect to the fix South China block. Thick black lines are
major active faults in the region. White line is approximate location of our finite element modeling profile

123
1264 Nat Hazards (2013) 69:1261–1279

The deep structure responsible for the Wenchuan earthquake is not known well because
the steep topography across the Longmen Shan prevents acquisition of high-resolution
seismic data. Surface exposures of the earthquake rupture along the Yingxiu-Beichuan
fault show dip angles of 70°–80° to the northwest and 40°–60° to the northwest along the
Guanxian–Jiangyou fault (Li et al. 2008; Liu et al. 2008; Xu et al. 2008a, b, 2009).
Through waveform of fault zone trapped wave analysis, Li et al. (2009) found that the fault
zone must dip *70° northwestward to a depth of 10 km. At the same time, joint inversion
of GPS and InSAR data by Shen et al. (2009) shows that high-angle oblique reverse
faulting is required to produce the geodetically observed co-seismic displacement field.
On the other hand, fault plane solutions of the mainshock indicate slip on a fault dipping
308 to 408 (CENC 2008; Ji 2008; Nishimura and Yagi, 2008; Zhang et al. 2008a, b; The
Harvard CMT catalog 2008). The precisely relocated epicenter of the mainshock is only
8 km southwest of the primary surface rupture at depth about 18.7 km, the Yingxiu–
Beichuan fault (Chen et al. 2009). If the earthquake rupture took place on a planar fault
dipping 30° or 40°, for the given mainshock hypocenter location, the surface rupture would
appear 22–32 km east of the Yingxiu-Shuiguan rupture and within the Sichuan basin.
Based on arguments above, Zhang et al. (2009a, 2010a) proposed that the structure
responsible for the Wenchuan earthquake consists of imbricate, oblique, high-angle listric
reverse faults. The faults dip *70° above 15 km depth, become 30° to 40° below *15 km
depth, and presumably root into sub-horizontal brittle–ductile transition zone below
22 ± 2 km depth. According to various fracture criteria, slip is easy on a fault dipping
30°–40° under the stress regime of pure shear. The initial slip on the gentle-dipping fault
probably has caused the Coulomb stress changes that may in turn trigger significant slip on the
high-angle dipping fault above it to form the Wenchuan earthquake (Zhang et al. 2010a).

3 Model setup

In order to examine the hypothesis mentioned above and to understand physical processes
associated with strain buildup and release during the inter- and coseismic phases of the
Wenchuan earthquake, a viscoelastic finite element model is constructed to simulate cycles of
earthquake generation and inter- and co-seismic behaviors of the Longmen Shan fault zone.
Therefore, a vertical section crossing the epicenter of the main shock and perpendicular to the
strike of the Longmen Shan fault is chosen as our model plane shown in Fig. 1. The model is
350 km deep and 600 km long to include part of the Sichuan basin and the Eastern Tibet. In
fact, about one half of the model domain is in Tibet, and the other half in the Sichuan basin.
The Moho depth in the left side of the model, the Eastern Tibet, is *65 km, and it is
*40 km deep in right side of the model, the Sichuan basin (Wang et al. 2007; Liu et al. 2009;
Zhang et al. 2009b, 2010b). But the Moho beneath the Longmen Shan fault is assigned to
about 60 km according to seismic profiling result made by Zhang et al. (2009b). The dividing
line between the upper crust and the lower crust in the Eastern Tibet is assumed as a brittle–
ductile transition below which rheological deformation may occur (Zhang et al. 2010a).
Numerous geological and geophysical studies indicate remarkable contrasts in litho-
spheric structure among the Eastern Tibet, the Longmen Shan, and the Sichuan basin (Bur-
chfiel et al. 1995, 2008; Royden et al. 1997, 2008; Wang et al. 2007; Yao et al. 2008; Liu et al.
2009; Wang et al. 2009; Zhang et al. 2009b, 2010b). The model is set up with different
rheological structures for Sichuan basin, the Longmen Shan, and the Eastern Tibet.
The whole crust in the Sichuan basin (denoted by Sc in Fig. 2) is regarded as elastic due to
its minor surface deformation, low level of seismicity, and rapid seismic wave velocities.

123
Nat Hazards (2013) 69:1261–1279 1265

Fig. 2 Geometry and boundary conditions of the finite element model. The displacements shown by black
arrows along x-axis are applied at each node on the left side of the model, with its value consistent with GPS
velocity vectors. The model is divided into nine sub-regions based on previous geological and geophysical
studies. Sc denotes whole crust of Sichuan Basin, and Sm denotes the mantle beneath Sichuan basin. LMS is
the upper crust of Longmen Shan. Tuc1, Tlc1, and Tm1 represent upper crust, lower crust, and upper mantle
of the West Longmen Shan. Tuc2, Tlc2, and Tm2 denote upper crust, lower crust, and upper mantle of the
region west of the Longriba fault, the Songpan Terrain

The mantle beneath the Sichuan basin (Sm in Fig. 2) is assumed to be ‘‘strong’’ as suggested
from seismic tomography (e.g., Li and van der Hilst 2010). The upper crust of Longmen Shan
(Lms in Fig. 2) is set to be very strong and difficult to deform, but its lower crust and upper
mantle are as ‘‘weak’’ as that of the Eastern Tibet (Liu et al. 2009; Wu et al. 2009). Seismic
tomography and receiver function inversion suggests that lithospheric structure of the Eastern
Tibet appears to be different across the Longriba fault, about 200 km west of the Yingxiu–
Beichuan fault (Liu et al. 2009; Wu et al. 2009). The region between the Longriba fault and the
Longmen Shan is defined as West Longmen Shan that is characterized by a seismically slow
lithospheric structure which is ‘‘soft’’ upper crust (Tuc1 in Fig. 2), lower crust (Tlc1 in
Fig. 2), and upper mantle (Tm1). The region west of the Longriba fault (Songpan Terrain)
also has ‘‘soft’’ lithospheric structure (upper crust Tuc2, lower crust Tlc2, and upper mantle
Tm2 in Fig. 2) relative to structure of the Sichuan basin, but its strength appears to be a little
more ‘‘strong’’ than the West Longmen Shan.
In order to obtain the best-fitting model during interseismic period, Zhu and Zhang
(2012) performed inversion by means of grid search to find the optimum model parameters,
including Young’s modulus, Poisson’s ratio, and viscosity. The inverted optimal param-
eters for each region in Fig. 2 are listed in Table 1. Therefore, simulation in the paper is
conducted with the model parameters chosen according to Table 1.
The geometry of meshes in the model is also shown in Fig. 2, in which there are 8,630
triangular 3-node elements connected by 4,432 nodes.
In the finite element calculation, fault behavior, being in the state of stick–slip, is
modeled by means of contact element method (Zhu and Zhang 2009, 2010, 2012). A
sudden slip on fault is assumed as an earthquake event. On the contrary, in the period of

123
1266 Nat Hazards (2013) 69:1261–1279

Table 1 Model material parameters used in modeling

Region Parameter Description Value

Tuc1 Euc Young’s modulus 6.5 9 1010pa

luc Poisson’s ratio 0.26
Tuc2 Euc Young’s modulus 6.8 9 1010pa
luc Poisson’s ratio 0.26
quc Density of upper crust 2.6 Mg m-3
Tlc1 Elc Young’s modulus 6.3 9 1010pa
llc Poisson’s ratio 0.30
vc Viscosity of lower crust 4.5 9 1018pa
s
Tlc2 Elc Young’s modulus 6.5 9 1010pa
llc Poisson’s ratio 0.27
vc Viscosity of lower crust 6.5 9 1018pa
s
qlc Density of lower crust 2.65 Mg m-3
Sc Esc Young’s modulus 3.3 9 1011pa
qsc Density of Sichuan crust 2.7 Mg m-3
lsc Poisson’s ratio 0.25
Tm1 Em Young’s modulus 7.6 9 1010pa
lm Poisson’s ratio 0.26
vm Viscosity of upper mantle 6 9 1020pa
s
Tm2 Em Young’s modulus 7.8 9 1010pa
lm Poisson’s ratio 0.26
vm Viscosity of upper mantle 3 9 1021pa
s
Sm Esm Young’s modulus 3.6 9 1011pa
lsm Poisson’s ratio 0.25
vm Viscosity of upper mantle 3 9 1022pa
s
qm Density of upper mantle 3.2 Mg m-3
Lms Ep Young’s modulus 7.8 9 1010pa
lp Poisson’s ratio 0.23
qp Density of Longmen Shan 2.7 Mg m-3
All area g Gravitational acceleration 9.8 m s-2
l0 Coefficient of friction 0.60

Names of the regions are corresponding to those in Fig. 2

interseismic, fault is in the state of locking, with some strains accumulated on it. Friction
on fault surface poses important impact on kinematic and seismic behavior along a locking
seismogenic fault. When the fiction coefficient increases, the frictional strength of the fault
also increases so that the locking fault could store large amount of elastic strain energy to
generate great events like the 2008 Wenchuan earthquake. The friction on fault zone is
very complex (Tse and Rice 1986; Dieterich 1994; Scholz 1990, 1998). For simplicity, we
adopt the rate-dependent friction constitutive relation (Zhu and Zhang 2010). When
modeling fault behaviors by contact element, a problem of convergence in calculation is
usually confronted. Contact with friction is a highly nonlinear problem in finite element
method.
How to apply boundary conditions is one of the key problems in finite element mod-
eling. Based on geological studies and GPS measurements (Shen et al. 2005; Gan et al.

123
Nat Hazards (2013) 69:1261–1279 1267

2007; Zhang et al. 2008a), the boundary conditions are applied as follows: The surface of
the model is set to free, the right side of the model is set to zero in horizontal direction and
is free vertically (shown in Fig. 2), because the Sichuan basin is regarded as stable rigid
block. The bottom of the model is free to move in the horizontal direction and is fixed to
zero vertically. The left side is applied displacements, with the magnitude obtained
according to the speed of 5 mm/yr, the average value of the GPS velocity vectors, with
directions along x-axis, pointing to the Sichuan basin (Fig. 2).
The initial stress condition is very important for FEM simulation, but it is very difficult
to obtain initial stress state in deep earth. We have to rely on artificial assumptions that
combination of gravity and boundary condition forms initial stress field (Zhu and Zhang
2010, 2012). On the base of initial model, boundary displacements are applied based on
calculation time, with the magnitude in accordance with the velocity surveyed by GPS.
In modeling fault behavior, some parameters such as friction coefficient are assigned by
calling Fortran subroutine UFRIC in commercial software MSC
MARC, which can be
modified by any user. In the paper, calculation is performed by quasi-static state problem
with 60,000 computational steps.

4 Model results

A three-step modeling is utilized to simulate inter-, co-seismic deformation, and slip

history on the fault. The first step is to simulate interseismic deformation before the 2008
Wenchuan earthquake. Then, the model continues until a seismic event occurs, and to
simulate the coseismic displacements of the Wenchuan earthquake. At last, the model will
yield slips on the fault one after another to simulate seismicity around the Longmen Shan
fault under the loading of continuous constant boundary conditions. It should be noted that
different computational time step is adopted in different period of time. For example, in
coseismic period, the time step is very short, being about 0.01 s, but it is long in
interseismic period with the value of 107 s.

4.1 Interseismic deformation

4.1.1 Interseismic strain rates

Using the model parameters described above, the interseismic strain rate is computed
across the Longmen Shan. Figure 3 displays the maximum shear strain rate distribution.
The most striking features of the strain rates are the concentration of strains in the lower
crust of the West Longmen Shan region and almost negligible strain accumulation in the
lithosphere of the Sichuan basin (Fig. 3). On the surface, the calculated strain rates are over
10 times less than the maximum values in the lower crust, but the largest strain rates still
present west of the seismogenic fault for about 40–50 km in length.
It is noticeable that strain rates along the listric seismogenic Yingxiu–Beichuan fault has
been very slow as shown in Fig. 3 that surface exposure of the fault locates at 0 km in
horizontal coordinate, dips toward the Eastern Tibet, and extends downward for about
25 km at depth. Also, the maximum shear strain rates in the Sichuan basin are very small
with the magnitude less than 10 3 10-9/yr. This is in good agreement with observations
that the interseismic deformation across the Longmen Shan fault has been very slow or
almost negligible (Burchfiel et al. 1995, 2008; Densmore et al. 2007; Zhang et al. 2010a),
implying intensive seismic locking before the 2008 Wenchuan earthquake.

123
1268 Nat Hazards (2013) 69:1261–1279

Fig. 3 Map of the maximum shear strain rate in interseismic phase before the 2008 Wenchuan earthquake.
LMS stands for the Longmen Shan. Pink curve denotes Longmen Shan thrust fault that dips *70° in the
upper part and dips only 30° in the lower section

4.1.2 Elastic strain energy density

Elastic strain energy accumulation is proportional to seismic potential. Distribution of

elastic strain energy density rates (ESED) is shown in Fig. 4. The figure clearly shows
significant concentration of ESED on the hanging wall of the listric Yingxiu–Beichuan
fault with the maximum value of more than 200 J/m3/yr, which may be reversible in next
earthquakes. In particular, the ESED accumulates very rapidly in and around the Longmen
Shan fault in the depth above *25 km, whereas it decreases dramatically below 30 km
depth beneath the Longmen Shan. The low ESED zone extends downward and roots into
the lower crust of the Eastern Tibet where the ESED is almost zero (Fig. 4). Except its
western edge, the entire Sichuan basin shows very slow or no ESED accumulation which is
consistent with the observed low level of seismicity and absence of Holocene active
faulting. Upper crust of the Eastern Tibet is characterized by moderate to high level of
ESED accumulation.
The rapid ESED on the Longmen Shan fault implies that the fault as a seismic locked
zone has a capability to store significant elastic strain energy to generate great earthquake
even if the rate of deformation is slow on the surface, and explains why the great
Wenchuan earthquake occurs on the region of such a slow surface deformation.

4.1.3 Sensitivity of some model parameters

The modeling results indicate that the rheological contrasts of lithospheric structure across
the Longmen Shan fault dictate pattern of distribution of elastic strain energy density rates.
In order to attest the notion, sensitive tests of crustal material properties are conducted
(Fig. 5). One of the tests is to replace the ‘‘strong’’ upper crust of the Longmen Shan (Lm)
with a ‘‘soft’’ upper crust of the West Longmen Shan (Tcu1). The pattern of ESED dis-
tribution changes dramatically (Figs. 4, 5a) that the rapid accumulation of ESED only
takes places near the surface of the Yingxiu–Beichuan fault for only 8–10 km depth. ESED
accumulation in such a shallow and localized region is unlikely store energy as large as
that released by the 2008 Wenchuan earthquake. Thus, existence of the ‘‘strong’’ upper

123
Nat Hazards (2013) 69:1261–1279 1269

Fig. 4 Distribution of elastic strain energy density rate (ESED) in interseismic period before the 2008
Wenchuan earthquake. The largest value is located in and around the Yingxiu–Beichuan fault, with the
lowest in the Sichuan basin and in the lower crust of the Eastern Tibet. It is the rapid accumulation of the
elastic strain energy in Longmen Shan region that leads to the occurrence of the 2008 Wenchuan earthquake

crust of Longmen Shan is necessary for energy accumulation and devastating earthquake
generation. Model tests also show that a ‘‘weak’’ lower crust beneath the entire Eastern
Tibet is a prerequisite for energy accumulation and great earthquake occurrence. Without
the ‘‘weak’’ lower crust beneath the entire Eastern Tibet (Fig. 5b), rapid ESED accumu-
lation would occur on lower crust between 30 and 60 km depth range along the Yingxiu–
Beichuan fault and there would be no rapid ESED accumulation on upper crust of the
Longmen Shan which was the place of major coseismic energy release during the 2008
Wenchuan earthquake (e.g., Nishimura and Yagi 2008; Ji and Hayes 2008; Zhang et al.
2009a, b; Wang et al. 2011). In addition, the rigid Sichuan basin lithosphere is very
important for energy accumulation in the sensitive test. Without very ‘‘strong’’ lithosphere
of the Sichuan basin (Table 1), there will be no rapid ESED accumulation in the Longmen
Shan regardless whether ‘‘strong’’ or ‘‘weak’’ upper crust of Longmen Shan exists or not
(Fig. 5c).
In sum, it is concluded that the rigid Sichuan basin, ‘‘soft’’ lower crust beneath the
Eastern Tibet and ‘‘strong’’ upper crust of the Longmen Shan altogether contribute the
rapid elastic strain energy accumulation in and around the Longmen Shan fault zone.

4.2 Co-seismic slip distribution

On the base of the interseismic deformation model above, boundary displacements are
continued to impose until large earthquake occurs. In order to study detailed physical
processes immediately before and after the main earthquake rupture, the time step for
computation is reduced to 0.01 s. Figure 6 displays coseismic displacement distribution
across the profile through forward modeling using finite element model. The figure shows
that general pattern of co-seismic displacements roughly agrees with observations
(Xu et al. 2008a, b, 2009; Shen et al. 2009; Zhang et al. 2010a) although the modeled
values are a litter larger than the observed displacements. Most of deformation occurred
within *30 km range around the seismogenic fault with *20 km in hanging wall and

123
1270 Nat Hazards (2013) 69:1261–1279

Fig. 5 Elastic strain energy density rate (ESED) distribution in interseismic phase with some model
parameters changed. a ‘‘soft’’ Longmen Shan model, where the upper crust of Longmen Shan is replaced by
the material with the properties of region Tuc1. b The ESED distribution where ‘‘soft’’ lower crust on the
Eastern Tibet is replaced by overlying upper crust while the ‘‘strong’’ Longmen Shan remains. c The rigid
Sichuan basin is replaced by ‘‘soft’’ material of Eastern Tibet (Tuc1) whether the strong Longmen Shan
remains or not in the model

123
Nat Hazards (2013) 69:1261–1279 1271

*10 km in the footwall. Vertical slips on the hanging wall are larger than those on
footwall, whereas horizontal shortenings on the hanging wall are less than those on the
footwall (Fig. 6).
The pattern of coseismic slip vectors expresses a typical reverse faulting slip distri-
bution, with slips on the hanging wall moving upwards and slips on the footwall going
downwards (Fig. 6). The prominent feature of the simulated co-seismic slip is that the
vertical uplift is confined to narrowly distributed region within less than *20 km, which is
quite consistent with the pattern of aftershock distribution (Chen et al. 2009; Zhang et al.
2010a), surface rupture, and peak ground acceleration attenuation (Li et al. 2008; Wang
et al. 2009a, b; Zhang et al. 2010a; Wang et al. 2011) as well as other observations.

4.3 Slip history

One could speculate that slips on the surface of the fault will be produced one by one if
boundary loading is imposed continuously over a long period of time. Figure 7 shows
modeled accumulated slips at typical nodes on different segments of the modeled fault
within 25,000 years. As mentioned above, a sudden slip on the fault in simulation is
assumed as a seismic event, larger slip corresponds to a strong earthquake, and minor slip
to a small one. In Fig. 7, the black curve on the lower part of the figure is the slip history on
the steeply dipping segment of the fault; thick black bars indicate major slip events. The
gray line is the slip history of the gently dipping segment of the fault. Several small steps
are present on the gentle-dipping segment of the fault within each interval of major slips on
the steeply dipping segment of the fault. Thick gray bars are large slip events on the gentle-
dipping segment of the fault.
The most remarkable feature of the figure is that the recurrence intervals of strong
earthquakes on the fault are irregular although the boundary displacement rate is constant
with time. Eight modeled strong events (Fig. 7) and seven recurrent intervals can be
recognized from the modeled time-slip curves. The recurrent intervals of major events
range from 1843 to 4603 year with an average *3,300 years. Although the longest
interval is 2.5 times the shortest, 5 of the 7 modeled intervals are larger than 3,000 years.

Fig. 6 Computed coseismic displacements on the profile across the high-angle listric reverse Longmen
Shan fault (LMS). The scale of coseismic slip vector is shown in the low-right

123
1272 Nat Hazards (2013) 69:1261–1279

Fig. 7 Calculated time-slip history of the listric reverse fault during the first 25,000 year of modeling. Each
step in the curve represents a slip event on the fault. The black curve is slip history on the steeply dipping
segment of the fault, and thick black bars indicate major slip events. Grey line is slip history of gently
dipping segment of the fault. Thick grey bars are large slip events on the gentle-dipping segment of the fault.
Dashed grey lines show complying with slip-predictable model for small-slip events and with time-
predictable model for large events on the gently dipping segment. The numbers on the steps stand for
earthquake recurrence interval

The modeled average interval (3,300 year) roughly agree with that from geological,
geodetic, and seismological observations, and help to explain the absence of historical
major (M [ 7) earthquakes in the Longmen Shan region (Ran et al. 2008, 2010; Zhang
et al. 2008a; Wen et al. 2009).
If the slips are examined in depth, it is found that minor slips on the gently dipping
segment of the fault obey slip-predictable model (Shimazaki and Nakata 1980). The large
slips on the gently dipping fault comply with time-predictable model of recurrences. These
larger slips on the gently dipping fault occur simultaneously with slips on the steeply
dipping fault and may have triggered slips on the steeply dipping fault to form great events
such as the 2008 Wenchuan earthquake, whereas the major slips on the steeply dipping
segment of the fault do not obey any predictable model at all, indicating the strong events
on the fault behave more complicatedly.
On the other hand, from the analysis above, it could be inferred that frictional coeffi-
cient on surface of the fault will influence the slip history. Therefore, simulation is per-
formed with varied frictional coefficient, while other model parameters unchanged to
examine the sensitivity of the model. If the frictional coefficient is set to be as low as
l0 = 0.25, the steeply dipping segment slips more or less regularly with more than 20 m
displacement per event and accumulates much larger total displacement than that the
gently dipping segment (Fig. 8a). A huge slip of more than 200 m takes place after
20,000 years in our modeled slip history that is unlikely to happen from our understanding
of earthquake mechanics. The high frictional coefficient (l0 = 0.85) produced only less
than 4 m and less than 45 m total displacement on the steeply dipping and gently dipping
segments respectively for 35,000 years (Fig. 8b). Both low and high frictional coefficients
fail to simulate the characteristic slip behavior and 3300-year recurrent interval (Zhu and
Zhang 2010; Ran et al. 2008, 2010).

123
Nat Hazards (2013) 69:1261–1279 1273

Fig. 8 Calculated time-slip history of the listric reverse fault during the first 35,000 year of modeling, on a
listric reverse fault with 70° angle above 15 km depth, and 30° angle below 15 km depth. a The case of low
frictional coefficient and b is the case of high frictional coefficient. The red curve is slip history on the
steeply dipping segment of the fault, and green curve is slip history of gently dipping segment of the fault

5 Implications for earthquake prediction

5.1 Possible approach to earthquake prediction

From the modeling results above, it is clear that the region of the concentration of ESED is
a potential place for future earthquakes, which could be helpful to predict locus of future
events. At the same time, the next major slip (or strong event) on the Longmen Shan fault
can also be evaluated by means of numerical calculation. Thus, it can be said that time and
place of future earthquakes can be predicted to a certain extent by means of numerical
simulation. Of course, the accuracy of predicted time and space of future earthquakes
depends on whether the model parameters are chosen correctly or not. If model parameters,
such as material properties, friction laws on the fault, and initial stresses, are accurately
revealed by geophysical, geological, and geodetic methods, future earthquakes can be
precisely predicted only by numerical computation. The approach herein to predict
earthquakes by means of computation is called numerical earthquake prediction (NEP).
Theoretically, NEP can make accurate prediction because earthquakes are forecasted by
numerical calculation which is based on rigorous mathematical and physical principles,
rather than on empirical precursors or on statistics. Like weather forecast today, NEP will
be a bright avenue to successful earthquake prediction, playing an important role in future
natural hazard mitigation.
According to Fig. 7, one may notice that the slip history on the fault behaves irregular
occurrence pattern although the model in this study is very simple. This model behavior is
a typical characteristic of nonlinear system. However, it is found that main characteristics
of slip history are stable when the model is subjected to small perturbations, which imply
the model is robust.
As for nonlinear system, Geller et al. (1997), Geller (1997) claimed that the Earth is in a
state of self-organized criticality (SOC) where any small earthquake has some probability
of cascading into a large event. Whether or not this happens depends on unmeasured fine
details throughout a large volume, not just the immediate vicinity of the hypocenter. Thus,
earthquakes are probably inherently unpredictable.

123
1274 Nat Hazards (2013) 69:1261–1279

In reality, the SOC of earthquakes does not preclude useful predictions. Many other
natural hazards including floods, severe storms, wildfires, and climatic changes also exhibit
chaotic behavior. It has been examined that other natural processes such as volcanic
eruptions, severe storms, and climate change and something like that are also examples of
complex processes, each with its own predictable, possibly predictable and inherently
unpredictable elements. That a natural system is complex does not mean that predictions
are not possible for some scales of spatial, temporal, and magnitude.
Several aspects of the earthquake process indicate that it is less complex than most
aspects of the circulation of the atmosphere. Unlike the atmosphere, which moves sig-
nificantly even on short time scales, faults remain stationary over periods of tens of
thousands of years. Faults do not change their configuration significantly even over several
cycles of large events since the displacement in the largest shocks is at most meters to tens
of meters (Sykes et al. 1999). Atmospheric forecasts are made routinely for the next several
days. The success of weather forecast through numerical simulation demonstrates that
large earthquakes should be predicted by NEP as long as the physical processes of
earthquakes are totally understood. At least, long-term and perhaps intermediate-term
prediction for large earthquakes seems to be possible for very active fault segments at
present where large amount of data are available.
On the base of numerical simulation, Ben-Zion et al. (2003) assumed that large
earthquake cycles on heterogeneous faults are associated with intermittent criticality, not
always in SOC state, and large earthquake cycles may be tracked with seismicity functions,
suggesting earthquake prediction is possible.

5.2 Much complicated and difficult in earthquake prediction

As a physical phenomenon, earthquakes can be predictable to a certain degree. In par-

ticular, accurate successful earthquake will be achieved by means of NEP, theoretically.
However, the problem is difficult because the source volume inside the earth is inaccessible
to direct observation and because the most important parameters, the stress tensor, cannot
be measured directly (Wyss 2001). In addition, other numerous parameters in model are
not obtained directly. They are all obtained through geophysical inversions based on the
observation data on the earth surface. It is well known that geophysical inversion is not
unique. Usually several solutions can meet the observation data in inversion. Model
parameters like fault geometry, strength variations in the fault zone material, rheological
properties, state of stress, and flows of underground water are all the results inverted
through the surface observed data.
For example, using seismic p- and s- seismic velocity distributions, one can calculate
elastic constants of material properties such as Young’s modulus and Poisson’s ratio in the
crust. But, the resolution of seismic velocity is rather low; thus, they cannot meet the high
accuracy of earthquake prediction by NEP. Moreover, as Crampin (1999) pointed out in
the nature debate (http://www.nature.com/nature/debates/earthquake/) that earthquakes are
complex, they vary with magnitude and direction of stress field; shape of the fault planes;
orientation of fault plane with respect to stress field; presence or absence of fluids; nature
of fluids; fluid-pressure; asperities on fault plane; debris on fault plane; presence or absence
of fault gouge; presence or absence of water channels, pressure seals; height of water table;
temperature; state of Earth tides; state of ocean tides; air pressure; local geology; other
earthquakes; and so on. Each of these phenomena could in certain circumstances have
major effects on time, place, and magnitude of impending earthquakes. Consequently, no
two earthquakes are identical.

123
Nat Hazards (2013) 69:1261–1279 1275

Considering the above factors in simulation, earthquake prediction through NEP is

much complex and difficulty. Nevertheless, we could make earthquake prediction with
time window of *10 years at present for strong events under the circumstances in which
model parameters are evaluated based upon the current geosciences and technology in
theory. With the advancement of sciences, NEP is believed to become more and more
accurate.

6 Discussion and conclusion

Based on numerical simulation, this paper presents the ESED accumulation and slip history
on the Longmen Shan thrust fault by means of numerical simulation with viscoelastic finite
element method. The modeling results are consistent with the observed by paleoseismo-
logical investigations and the estimates by other methods, not only suggesting the method
applied in the study is reasonable, but also revealing significant applications in earthquake
prediction of the simulation results. In the paper, a conceptual model of NEP is proposed
on the base of numerical simulation. It is simple and initial, yet it did not make any realistic
earthquake prediction even in the Longmen Shan area. In order to enable NEP to make
progress in future, the following great efforts should be made.

6.1 Construction of observation networks

The model parameters in NEP are all deduced from observation data, which decide the
accuracy of NEP. So enough high-quality data are essential to NEP. Therefore, densely
deployed observation network is necessary to obtain all types of geophysical, geological,
and geodetic data. In some regions suitable for the research of NEP like Tibet, where
seismic activity is high and tectonics is complicated, we may deploy all types of obser-
vation stations. A favorable place for performing research of NEP may be Japan, where
large earthquakes occur frequently, and where GEONET, F-net, and Hi-net have been
working satisfactorily (Okada et al. 2004), which provides a large amount of data for NEP
research.
With a wealth of data from high-quality instruments, we can construct reasonable model
for NEP, and promote NEP to success finally.

6.2 Study on the physical processes of strong events

Predicting earthquakes requires an understanding of the underlying physics such as

nucleation of earthquakes, earthquake ruptures, and physical relation between precursory
phenomena and ensuing main shocks.
In simulation, stick–slip frictional behavior of fault (Scholz 1998; Zhu and Zhang 2009,
2010, 2012) based on numerical equations is particularly important. It has become possible
to simulate the complex and diverse phenomena which actually occur by creating
repeatedly occurring earthquakes hypothetically on a computer and adjusting the boundary
conditions for virtual reality. The role of fluids to strong earthquakes is worth studying
(Sibson 1992). How to simulate behaviors of underground fluids and how fluids trigger
large earthquake events should be paid special attention to. In addition to studying the
processes of large earthquakes on computer, laboratory simulation of earthquake’s pro-
cesses is important as well.

123
1276 Nat Hazards (2013) 69:1261–1279

6.3 Conduct research on NEP and put it to practice

It should be clear that what is impossible today does not mean what is impossible in future.
Therefore, that some scientists such as Geller believed earthquakes cannot be accurately
predicted at present does not demonstrate earthquakes cannot be forecast in future. NEP is
surely successful in the near future.
The key problem is that it needs a number of scientists doing research on NEP. Gov-
ernment should change their mind from searching seismic precursors to support NEP
research. In this regard, China set us a good example and funded many research programs
to study earthquake prediction. Moreover, China has founded a special government
administration named ‘‘China Earthquake Administration (CEA).’’ The main task of CEA
is to monitor precursors and to do research on earthquake prediction.
NEP is very complicated and much difficult. We must exercise patience and do not
expect spectacular success quickly. At first, we could do some prediction attempts in a
single fault such as the Xiangshuihe fault in the Eastern Tibet, where large earthquakes
occur frequently and a wealth of geophysical and geological data are available. Then, step
by step, we carry out complex NEP research on large scale in space and time.
Based upon the model simulation in the paper and the analyses above, some primary
conclusions are as follows.
Physical processes of the Wenchuan earthquake could be well simulated by means of
viscoelastic finite method. The computed strain rates accumulate much faster in the lower
crust of the Eastern Tibet than those in other regions during interseismic period. At the
same time, the rapid ESED is located in and around the Longmen Shan fault zone, which
plays an important role in generation of the Wenchuan earthquake. This may explain the
question why the great Wenchuan earthquake occurred on the Longmen Shan fault with
small surface deformation. The modeled average recurrence interval of the strong earth-
quakes in the Longmen Shan fault is *3300 yr, which is in good agreement with the
observed by paleoseismological investigations and estimates by other methods.
From the modeling results in the paper, the time and place of the future earthquakes can
be evaluated by numerical simulation. NEP is a promising method to successful earthquake
prediction.
NEP, as a scientific discipline, is still in its infancy, and it needs improving to make
realistic earthquake prediction. Also, it needs paying attentions specially and well funding.
Although NEP is very much complicated and difficult, it is possible that NEP will play an
important part in future natural hazard mitigation.

Acknowledgments I greatly appreciate two anonymous reviewers and editor for their constructive
comments and suggestions on the present work. I also thank Prof. Peizhen Zhang and Prof. Zhengkang Shen
for their instructive suggestions. This work was supported by the National Natural Science Foundation of
China (No. 40974020), by the State Key Laboratory of Earthquake Dynamics (LED2012B01), and by
Special project from Ministry of Land and Resources of China (SinoProbe-07).

References

Ben-Zion Y, Eneva M, Liu Y (2003) Large earthquake cycles and intermittent criticality on heterogeneous
faults due to evolving stress and seismicity. J Geophys Res 108(B6). doi:10.1029/2002JB002121
Burchfiel BC, Chen Z, Liu Y, Royden LH (1995) Tectonics of the Longmen Shan and adjacent regions. Int
Geol Rev 37:661–735

123
Nat Hazards (2013) 69:1261–1279 1277

Burchfiel BC, Royden LH, van der Hilst RD, Hager BH, Chen Z, King RW, Li C, Lü J, Yao H, Kirby E
(2008) A geological and geophysical context for the Wenchuan earthquake of 12 May 2008, Sichuan,
People’s Republic of China. GSA Today 18:4–11. doi:10.1130/GSATG18A.1
CENC (China Earthquake Networks Center) (2008) http://www.csi.ac.cn/sichuan/index080512001.html
Chen Z, Burchfiel BC, Liu Y, King RW, Royden LH, Tang W, Wang E, Zhao J, Zhang X (2000) Global
positioning system measurements from Eastern Tibet and their implications for India/Eurasia inter-
continental deformation. J Geophys Res 105:16215–16227
Chen J, Liu Q, Li S, Guo B, Li Y, Qi S (2009) Seismotectonics study by relocation of the Wenchuan Ms8.0
earthquake sequence. Chinese J Geophys 52:390–397 (in Chinese with an English abstract)
Crampin (1999) http://www.nature.com/nature/debates/earthquake/
Densmore AL, Ellis MA, Li Y, Zhou R, Hancock GS, Richardson N (2007) Active tectonics of the Beichuan
and Pengguan faults at the eastern margin of the Tibetan Plateau. Tectonics 26:TC4005. doi:10.1029/
2006TC001987
Dieterich JH (1994) A constitutive law for the rate of earthquake production and its application to earth-
quake clustering. J Geophys Res 99:2601–2618
Gan W, Zhang P, Shen Z-K, Niu Z, Wang M, Wan Y, Zhou D, Cheng J (2007) Present-day crustal motion
within the Tibetan Plateau inferred from GPS measurements. J Geophys Res 112:B08416. doi:10.1029/
2005JB004120
Geller RJ (1997) Earthquake prediction: a critical review. Geophys J Int 131:425–450
Geller RJ, Jackson DD, Kagan YY, Mulargia F (1997) Earthquakes cannot be predicted. Science
275:1616–1617
Hubbard J, Shaw JH (2009) Uplift of the Longmen Shan and Tibetan plateau, and the 2008 Wenchuan
(M = 7.9) earthquake. Nature 458:194–197. doi:10.1038/nature07837
Ji C (2008) Preliminary result of the May 12, 2008 MW 7.9 eastern Sichuan, China earthquake, May 12,
2008: http://www.geol.ucsb.edu/faculty/ji/big_earthquakes/2008/05/12/ShiChuan.html
Ji C, Hayes G (2008) Preliminary result of the May 12, 2008 Mw 7.9 Eastern Sichuan, China earthquake.
http://earthquake.usgs.gov/eqcenter/eqinthenews/2008/us2008ryan/finite_fault.php2008
Kagan YY (1997) Are earthquake predictable? Geophys J Int 131:505–525
King RW, Shen F, Burchfiel BC, Royden LH, Wang E, Chen Z, Liu Y, Zhang X-Y, Zhao J-X, Li Y (1997)
Geodetic measurement of crustal motion in southwest China. Geology 25:179–182
Kirby E, Whipple K, Harkins N (2008) Topography reveals seismic hazard. Nat Geosci 1:1–3
Li C, van der Hilst R (2010) Structure of the upper mantle and transition zone beneath Southeast Asia from
traveltime tomography. J Geophys Res 115:B07308. doi:10.1029/2009JB006882
Li H, Fu X, van der Woerd J (2008) Surface rupture associated with the Wenchuan earthquake and its
oblique slip. Acta Geol Sinica 82:1623–1643
Li S, Lai X, Yao Z, Yang Q (2009) Study on fault zone structures of northern and southern portions of the
main central fault generated by Ms = 8.0 Wenchuan earthquake using fault zone trapped waves.
Earthq Sci 22(4):417–424. doi: 10.1007/s11589-009-0417-3
Liu J, Zhang Zh, Wen L (2008) The Ms 8.0 Wenchuan earthquake co-seismic rupture and its tectonic
implications—an out-of-sequence thrusting event with slip partitioned on multiple faults. Acta Geol
Sinica 82:1707–1722 (in Chinese with an English abstract)
Liu Q, Li Y, Chen J, Guo B, Li S, Wang J, Zhang X, Qi S (2009) Crustal and mental shear wave velocity of
the Eastern Tibetan Plateau and Sichuan basin: implications for velocity structure of the 2008
Wenchuan earthquake. Chinese J. Geophys. 52:309–319 (in Chinese with an English abstract)
MSC
Software (2008) www.mscsoftware.com
Nature Debate on Earthquake Prediction, www.nature.com/nature/debates/earthquake, on the internet. Feb
1999
Nishimura N, Yagi Y (2008) Rupture process for May 12, 2008 Sichuan earthquake (preliminary result),
available at http://www.geol.tsukuba.ac.jp/nisimura/20080512/
Okada Y, Kasahara K, Hori S, Obara K, Sekiguchi S, Fujiwara H, Yamamoto A (2004) Recent progress of
seismic observation networks in Japan—Hi-net, F-net, K-NET and KiK-net. Earth Planets Space
56:15–28
Ran Y, Chen L, Chen G, Yin J, Chen J, Gong H, Shi X, Li C (2008) Primary analysis of in situ recurrence of
large earthquake along seismogenic fault of the Ms 8.0 Wenchuan Earthquake, China. Seismol Geol
30:630–643 (in Chinese with English abstract)
Ran Y et al (2010) Paleoseismic evidence and repeat time of large earthquakes at three sites along the
Longmenshan fault zone. Tectonophysics. doi:10.1016/j.tecto.2010.01.009
Royden LH, Burchfiel BC, King RW, Wang E, Chen Z et al (1997) Surface deformation and lower crustal
flow in Eastern Tibet. Science 276:788–790

123
1278 Nat Hazards (2013) 69:1261–1279

Royden L, Burchfiel BC, Van der Hilst RD (2008) The geological evolution of the Tibetan Plateau. Science
321:1054–1058
Scholz CH (1990) The mechanics of earthquakes and faulting. Cambridge University Press, New York
Scholz CH (1998) Earthquakes and friction laws. Nature 391(1):36–42
Shen ZK, Lu J, Wang M, Burgmann R (2005) Contemporary crustal deformation around the southeast
borderland of the Tibetan Plateau. J Geophys Res 110:B11409. doi:10.1029/2004JB003421
Shen ZK, Sun J, Zhang PZ, Wan Y, Wang M et al (2009) A rare cascade rupture of a complex fault system
in the 12 May 2008 Wenchuan earthquake. Nature Geosci 2. doi:10.1038/NGEO636
Shimazaki K, Nakata T (1980) Time-predictable recurrence model for large earthquakes. Geophys Res Lett
7(4):279–282. doi:10.1029/GL007i004p00279
Sibson RH (1992) Implications of fault-valve behavior for rupture nucleation and recurrence. In: Mikumo T
et al. (eds) Earthquake source physics and earthquake precursors. Tectonophysics 211:283–293
Sykes L, Shaw B, Scholz C (1999) Rethinking earthquake prediction. Pure Appl Geophys 155:207–232
The Harvard CMT catalog http://www.globalcmt.org/CMTsearch.html
Tse S, Rice J (1986) Crustal earthquake instability in relation to the depth variation of friction slip prop-
erties. J Geophys Res 91:9452–9472
Wang CY, Han WB, Wu JP (2007) Crustal structure beneath the eastern margin of the Tibetan plateau and
its tectonic implications. J Geophys Res 112. doi:10.1029/2005JB003873
Wang CY, Mooney WD, Ding ZF, Yang JS, Yang ZX et al (2009) Shallow seismic structure of Kunlun fault
zone in northern Tibetan Plateau, China: implications for the 2001 Ms8.1 Kunlun earthquake. Geophys
J Int 178. doi:10.1111/j.1365-246X.2009.0409.x
Wang QL et al (2009b) Coseismic vertical deformation of the MS8.0 Wenchuan earthquake from repeated
leveling and its constraint on listric fault geometry. Earthq Sci 22:595–602
Wang Q, Qiao X, Lan Q, Jeffrey F, Yang S, Xu C, Yang Y, You X, Tan K, Chen G (2011) Rupture of deep
faults in the 2008Wenchuan earthquake and uplift of the Longmen Shan. Nat Geosci 4:634–640.
doi:10.1038/NGEO1210
Wen XZ, Ma SL, Xu XW, He YN (2008) Historic pattern and behavior of earthquake ruptures along the
eastern boundary of the Sichuan–Yunnan fault block, southwestern China. Phys Earth Planet Inter
168:16–36
Wen XZ, Zhang PZ, Du F, Long F (2009) The background of historical and modern seismic activities of the
occurrence of the 2008 Ms 8.0 Wenchuan, Sichuan, earthquake. Chinese J Geophys 52(2):444–454
Wu J, Huang Y, Zhang T, Ming Y, Fang L (2009) Aftershock distribution of the MS8.0 Wenchuan
earthquake and three dimensional P-wave velocity structure in and around source region. Chinese J
Geophys 52(2):320–328
Wyss M (2001) Why is earthquake prediction research not progressing faster? Tectonophysics
338(2001):213–217
Xu X, Wen X, Ye J, Ma B, Chen J et al (2008a) The Ms 8.0 Wenchuan earthquake surface ruptures and its
seismogenic structure. Seismol Geol 30:597–629 (in Chinese with an English abstract)
Xu Z, Ji S, Li H (2008b) Uplift of the Longmen shang range and the Wenchuan earthquake. Episodes
31:291–301
Xu X, Wen X, Yu G, Chen G, Klinger Y, Hubbad J, Shaw J (2009) Coseismic reverse- and oblique-slip
surface faulting generated by the 2008 Mw 7.9 Wenchuan earthquake, China. Geology 37:515–518
Yao H, Beghein C, van der Hilst RD (2008) Surface wave array tomography in SE Tibet from ambient
seismic noise and two-station analysis—II Crustal and upper-mantle structure. Geophys J Int
173:205–219. doi:10.1111/j.1365-246X.2007.03696.x
Zhang P, Shen Z, Wang M, Gan W, Burgmann R, Molnar P, Wang Q, Niu Z, Sun J, Wu J, Sun H, You X
(2004) Continuous deformation of the Tibetan Plateau from global positioning system data. Geology
32:809–812
Zhang PZ, Xu XW, Wen XZ, Rang RK (2008a) Slip rates and recurrence intervals of the Longmen Shan
active fault zone and tectonic implications for the mechanism of the May 12 Wenchuan earthquake,
2008, Sichuan, China. Chinese J Geophys 51:1066–1073 (in Chinese with an English abstract)
Zhang Y, Feng WP, Xu LS, Chen Y (2008) Temporal-spatial rupture process of Ms8.0 Wenchuan earth-
quake of 2008. Sci China Ser D Earth Sc 38. doi:10.1007/s11430-008-0148-7
Zhang PZ, Wen X, Xu X, Gan W, Wang M et al (2009a) Model of strain accumulation and release
associated with the 2008 Wenchuan, Sichuan, China earthquake. Chinese Sci Bull 54:944–953 (in
Chinese)
Zhang Zh, Wang YH, Chen Y, Tian XB, Houseman G et al (2009b) Crustal structure across Longmenshan
fault belt from passive source seismic profiling. Geophys Res Lett 36:L17310. doi:10.1029/
2009GL039580

123
Nat Hazards (2013) 69:1261–1279 1279

Zhang PZ, Wen X, Shen Z, Chen J (2010a) Oblique high-angle listric-reverse faulting and associated
straining processes: the Wenchuan earthquake of 12 May 2008, Sichuan, China. Annu Rev Earth
Planet Sci 38:353–382
Zhang Zh, Yuan X, Chen Y, Tian X, Kind R et al (2010b) Seismic signature of the collision between the east
Tibetan escape flow and the Sichuan Basin. Earth Planet Sci Lett 292:254–264
Zhou RJ, Li Y, Densmore AL, Ellis MA, He YL, Li YZ, Li XG (2007) Active tectonics of the Longmen
Shan region on the Eastern Margin of the Tibetan Plateau. Acta Geologica sinica (English edition)
81(4):593–604
Zhu SB, Zhang PZ (2009) A study on the dynamical mechanisms of the Wenchuan Ms 8.0 earthquake.
Chinese J Geophys 52:418–427 (in Chinese)
Zhu SB, Zhang PZ (2010) Numeric modeling of the strain accumulation and release of the 2008 Wenchuan,
Sichuan, China, Earthquake. Bull Seismol Soc Am 100:2825–2839
Zhu SB, Zhang PZ (2012) FEM simulation of interseismic and coseismic deformation associated with the
2008 Wenchuan earthquake, Tectonophysics. doi:10.1016/j.tecto.2012.06.024
Zhu SB, Xing HL, Xie FR, Shi YL (2008) Simulation of earthquake processes by finite element method: the
case of megathrust earthquakes on the Sumatra subduction zone. Chinese J Geophys 51:460–468

123