2000 AGU Fall Meeting
San Francisco, December 15-19, 2000


S71E-01 Invited
Review of Seismological Results from the 1999 Chichi, Taiwan, Earthquake
Mori, J., K.F. Ma


The September 21, 1999 Chichi, Taiwan earthquake (Mw=7.6) produced the presently best set of local and regional strong-motion records from a large earthquake. This shallow thrust event ruptured about 85 km of the Chelungpu fault in Central Taiwan with large displacements of 1 to 10 meters along most of the surface trace of the fault. The dense strong-motion array operated by the Seismology Center of the Central Weather Bureau recorded the earthquake on more than 400 stations, at distances of less than 1 km to 150 km from the fault. These data, along with locally recorded short-period data and teleseismic data from worldwide broadband stations provided valuable information about the earthquake source process, three-dimensional wave propagation, and regional tectonics. Using the strong-motion data, teleseismic data, and GPS displacement, several groups of researchers have produced models of the spatial and time distribution of slip and slip velocity for the earthquake. Common results of these studies show that slip occurred over an area of about 80 km by 40 km with the hypocenter in the southern region and rupture propagation mainly to the north. The seismic moment was 2 to 4 x10**27 dyne-cm. All of the models show a large asperity in the northern region of the fault with slip of over 10 m. This is in the region where very large surface displacments (8-10m) were measured on the fault. However, it difficult to model some of the strong-motion records very close to the fault, suggesting that some of the very large shallow slip did not radiate large amounts of elastic energy. Also, in the northern region of the fault where displacements are very large, the level of ground acceleration and associated building damage is relatively low. Estimates of the slip velocity show that the fault moved very rapidly (1-3 m/sec) but slip was "smooth" producing low levels of high-frequency radiation. There was a strong aftershock sequence with 7 events over M6, and several M5-6 events apparently triggered outside of the immediate region of the Chichi earthquake. Aftershocks distributions from permanent and temporary short-period networks show that most of the events are in the hanging wall of the fault and there are relatively few aftershocks along the shallow portion of the fault between the hypocenter (at about 7 km depth) and the surface trace of the fault. The aftershock pattern suggests that the Chelungpu fault is nearly horizontal at depths greater than 10 km, although there are large variations along strike. There are numerous clusters of aftershocks off the Chelungpu fault and a large number of events to the east, under the eastern edge of the Central Range. The pattern of aftershocks resembles the pattern of regional seismicity before the earthquake.



Rupture Process of the 1999 Chi-Chi, Taiwan, Earthquake From Strong Motion, Teleseismic and GPS Data
Ma, K., J. Mori, S. Lee, S., S.Yu


We investigated the rupture process of the 1999 Chi-Chi, Taiwan, earthquake, using high-quality near-source strong-motion records, broadband teleseismic displacement waveforms, and well-distributed GPS data. The near-source strong-motion displacement waveforms recorded significant static offsets of up to 8 m. The teleseismic displacement  records show a significant pulse with duration of about 18 to 20 sec. These data imply involvement of a large asperity during the source rupture. Considering the surface breaks observed along the Chelungpu fault, we considered two fault geometries: a single planar fault, and a two-segment fault with a northeast striking section near the northern end. Using the finite-fault model with various slip vectors, we derived the temporal and spatial slip distribution of the earthquake. The GPS data provided good surface displacement constraints in the slip distribution determination. The spatial slip distribution is generally consistent with field observations. The results, for the simple fault model show a large asperity located in the region about 25 km to 55 km north of the hypocenter with maximum slip of about 12m. The asperity further extends to the region where the fault bends to the northeast, which is consistent with GPS data, as found in the two-segment fault model. The slip amplitude near the hypocenter is about 3 to 6 m. The seismic moments determined from the various data sets are within the range of 2 to 4 x 10$^{27}$ dyne-cm. Most of the slip concentrated at the shallow depths (less than 10 km). The total rupture duration is about 28 sec and the rupture velocity is about 75\% to 80\% of the shear wave velocity. The slip vector shows a clockwise rotation during the fault rupture. The static stress drop of the large asperity region is comparable with the dynamic stress drop as observed directly from the slip velocity at the station near the large slip region.


Large Long-Period Earthquakes from Miyakejima Volcano Recorded at Kouzu Island
Mori, J., K. Sato, T. Shibutani


We studied several of the long-period events that are associated with eruptive activity from Miyakejima volcano using continuous strong-motion recordings from Kouzu Island and broadband Freesia data. We recorded about 15 large long-period events between July 1 and 14 at Kouzu Island, which is located about 40 km northwest of the volcano. At Kouzu Island, the seismograms of these events have long durations that last about 150 sec. The recordings have large amplitudes in the period range of 1 to 20 sec. Waveforms for the various long-period events look similar suggesting that they have similar mechanisms. Moment estimates for these events from the National Institute of Earth Science and Disaster Prevention, Science and Technology Agency are in the range of 10**23 to 10**24 dyne-cm (Mw 5.0 to 5.6). We carried out moment tensor inversions for several of these events using data in the period range of 10 to 100 sec. The waveforms are largely composed of P and S waves at Kouzu Island and surface waves at the farther Freesia stations. There were large non-double couple components in the moment tensor. The diagonal components of the moment tensor have similar values indicating a large isotropic component. The moment tensors can be interpreted as the sum of an isotropic component (expansion) and double couple of about equal size. The source time functions were also very long with durations of 10 to 40 sec. We estimate the depths to be about 2 km. Some of the long-period events occurred during times of eruptive activity at the volcano, while others occurred when the volcano was relatively quiet. The long time duration and isotropic component in the moment tensor suggest that the mechanism of these events involves movement



Estimates of Radiated Energy for Moderate Shallow Earthquakes in Japan
Kobayashi, H, J. Mori,  K.Sato


We estimated the seismic energy radiated by earthquakes in Japan using strong motion recordings from K-NET operated by the National Institute for Earth Science and Disaster Prevention, Science and Technology Agency. We analyzed data from 78 shallow events (depth less than 50 km) of magnitude 4.0 or greater that occurred from March 1997 through October 1999. Each event was recorded on 10 - 184 stations within 200 km of the epicenter. The method we used consisted of the time integration of the squared ground-motion velocity from the three components of ground motion, with a distance attenuation function and the station correction. The time window length is 50 sec from the first P-arrival.  We used a 0.1Hz highpass filter for the recordings. We separated rock and stiff soil sites from the other sites that had softer soil conditions and computed radiated energy for the two types of sites. Amplitudes for the soft sites averaged several times larger than for the hard sites. We consider the energy values from the hard sites to be a better estimate of the radiated energy. We computed station corrections by assuming that the average value of energy from the hard sites was correct, and calculated the average amplitude residual for each station using all the events. We also tested the effect of including the radiation pattern and using various attenuation relations. The results show that the log of the energy increases as a function of the log of the seismic moment with a slope slightly larger than 1.0. This relation is larger then expected for simple constant stress drop scaling and may indicate a trend in dynamic friction or driving stress as a function of earthquake size.


Strong Motion Recordings in the Izu Islands, Japan: Observations of Earthquake Initiations
Sato, K. and J. Mori


We recorded the strong earthquake swarm that was occurring during July and August 2000 in the Izu Islands on several strong-motion instruments deployed on Kouzu and Shikine Islands. We used 20 sec velocity sensors with both triggered and continous data loggers. The epicentral distances to the events were 5 to 40 km. We recorded on Kozu Island from July1 to 18 and on Kozu and Shikine Islands from August 9 to September 15. During these periods there were two M6, twelve M5, and numerous M4 events that were well recorded on our instruments. We use these data to look at details of the rupture initiation of moderate size events. For events recorded at close distances (less than 15 km) there are many clear waveforms that show the details of the beginning of the P-wave. We compare these observations to theoretical  results from crack models to estimate the size of the initiation zone.


Fault Plane Determination of the M7.6 Volcano Islands Earthquake
Kim, A and J. Mori


We determined the fault plane for the March 28, 2000 Volcano Islands earthquake (Mw7.6) in the Izu-Bonin subduction zone. This event at 100 km depth, is important for understanding the stresses within the down-going slab. The focal mechanism shows down-dip extension on either a nearly vertical or a nearly horizontal fault plane. If the nearly vertical nodal plane is identified as the actual fault plane, the slip along the fault plane is consistent with a sinking slab. On the other hand, if the orientation of fault plane is nearly horizontal, other stresses may be acting within the plate. To determine the actual fault plane, we analyzed the P and P depth phases at teleseismic distances (between 30 and 90 degrees) from 17 IRIS/GSN network stations. We used a least-squares inversion method to determine the slip distribution on the fault plane. We divided the fault plane into 135 subfaults and multiple time windows to solve for the slip. Various  rupture velocities from 3.0 to 4.5 km/sec were tested to see which rupture velocity and which fault plane produced the best fit to the data. It was difficult to resolve the rupture velocity. For the range of rupture velocites we tested, slip on the horizontal plane fit the data better than slip on the vertical plane. This result suggests that there are regional stresses near this kink in the Izu-Bonin trench that are conducive to horizontal faulting. Furthermore, in the Volcano Island region the Pacific plate is subducting obliquely beneath the overlying plate. This may cause bending in the slab which generates a complicated regional stress field.


2000 Seismological Society of Japan Fall Meeting
Tsukuba, November  20-22, 2000


Initiation of the Chichi, Taiwan Earthquake
Mori, J. and K.-f.



Strong motion recordings in the Izu Islands, Japan: Observations of Earthquake Initiations
Sato, K. and J. Mori


2000 Western Pacific Geophysics Meeting
Tokyo, June 27-30, 2000

Slip Velocity Estimates of the 1999 Chi-Chi, Taiwan, Earthquake: New Observations of Fault Dynamics
Mori, J and K.-F. Ma

We used the extensive near-field strong-motion data along with teleseismic data recorded for the 1999 Chichi, Taiwan earthquake (Mw 7.7) to study the distribution of slip velocity on the fault. Using three-component data from 23 near-field stations and P waves from 22 teleseismic stations we carried out a finite fault inversion to obtain the distribution of slip and slip velocity on the fault plane. The dense distribution of stations and close proximity to the fault allows us to make some of the best estimates of slip velocity ever done for a large earthquake. We find that the near the hypocenter the slip velocities are relatively low in the range of 0.3 to 0.7 m/sec. In the region of very large slip on the northern part of the fault we find very high slip velocities up to several meters per second. Since slip velocity is directly proportional to the dynamic stress drop, this region of large slip has a large dynamic stress drop. We suggest that as the rupture propagates toward the north, heat generated by the slip melts the fault zone or pressurizes fluids, which reducing the friction stress, increasing the slip velocity, and promoting the extensive sliding. Due to the dynamic rupture behavior such as melting or fluid pressurization, the large slip lubricates the fault zone, which is initially rather heterogeneous. This behavior smoothes the roughness of the fault zone, allowing a fast and smooth


Rupture Process of the 1999 Chi-Chi, Taiwan Earthquake From Direct Observations and Joint Inversion of Strong Motion, GPS and Teleseismic Data
Ma, K-F. and J. Mori

The strong motion stations along the surface ruptures of the fault directly recorded the dynamic  rupture behavior of the 1999 Chi-Chi (Mw7.7), Taiwan earthquake. The complete data set of this earthquake, including strong motion, GPS and teleseismic provide us with important information to understand the rupture process of the earthquake from several viewpoints. The observations from the near-fault strong motion data provides the most direct evidence on the rupture behavior of the fault. The joint inversion of the combined data provides insight into the temporal and spatial rupture process of the faulting. The results reveal the variations of spatial rotation of rake angle and slip velocity over the fault area, which implies the possible non-uniform pre-stress distribution and the evolvement of the rupture process. The GPS and near source strong motion data reveal similar slip distribution patterns; while the teleseismic data and velocity records of strong motion data show similar slip distribution patterns. These results imply that the large slip observed near the northern end of the fault is rather localized and might not radiate as much energy as expected. The large slip observed near the northern end of the fault has a large dynamic stress drop, equivalent to its static stress drop. We suggest that there may have been fault melting in this region to reduce the friction on the fault to very low levels.


Teleseismic and GPS Data Analysis of the 1999 Chi-Chi, Taiwan, Earthquake
Lee, S, K.-F.
Ma, J. Mori, S. Yu

We investigated 22 broadband teleseismic records and and well-recorded near-field GPS data to understand the rupture process of the 1999 Chi-Chi, Taiwan earthquake. The teleseismic waveforms are examined to determine the temporal and spatial slip distribution of the 1999 Chi-Chi, Taiwan earthquake. The analysis of the teleseismic waveform inversion through different models consistently reveals an anomalous large slip region centered about 40 to 50 km the north of the hypocenter at a depth from 0 to 10 km. The largest slip is about 6 - 10 m. The large slip region corresponds to the area where large surface displacements were observed on the fault and large displacements were measured from GPS observations. The slip near the vicinity of the hypocenter hypocenter is relatively smaller. The seismic- moment obtained from the teleseismic analysis is about 5-7x10**27 dyne-cm. The results also reveal significant variations of rake angle during the rupture. Some strike-slip component to the rupture was found in the middle portion of the fault. The slip distribution results from GPS data shows a larger slip region and larger amounts of slip compared with the teleseismic results. The difference between the GPS and teleseismic data might imply that the energy radiated by this earthquake is different for different frequencies. GPS data reflect the static displacements, which are not recorded by the teleseismic data. However, the overall patterns of the slip distribution and rake angle variations of these two data sets are quite comparable.


Drilling the Chelongpu Fault, Taiwan: Observations and Measurements of a Thrust-fault with Very Large Displacements and Velocities in a Recent Earthquake
Ando, M , J. Mori, H. Tanaka, H. Ishii, M. Yamano, C. Lee, K.-F. Ma, C. Wang, B. Huang, C. Lin, J. Wang, M.D. Zoback, R. Ikeda

We propose to drill the northern part of the Chelonpu fault to investigate the fault properties and rupture characteristics of the recent large Chi-chi, Taiwan (M7.7) earthquake of 1999. This earthquake presents the rare opportunity to probe an active fault that has just had large fault movements, up to 8 m of vertical surface displacement. Analyses of cores from the fault zone and heat flow measurements will provide new insights into the important physical properties, especially friction, that control the rupture of large earthquakes. Such direct observations of a recent large earthquake have never been made before. This project to drill the Chelongpu fault will lead to results in the following research areas. (1) Improved understanding of the whole fault zone structure and rock distribution for an active thrust-fault. (2) Estimates of fault surface velocity at the time of earthquake from core analyses. (3) Study of the mode of strain localization which results in high velocity motion and thus heat generation by mechanical wearing along the fault surface. (4) Study of fluid paths and permeability distributions in the fault zone. (5) Study of thermal properties and constraints on friction for the recent faulting process from heat-flow measurements. (6) High-quality seismic recordings of aftershocks from installation of borehole seismic and deformation instrumentation. (7) Study of earthquake triggering by carrying out a water-injection experiment.


Estimating Source Dimensions of the 1998 Earthquake Swarm in the Hida Mountains Region, Central Japan, Using Near-field Waveforms
Sato, K. and J. Mori

We estimated source durations for larger events in a strong swarm that occured from August 7 1998 in the Hida mountains region, central Japan. We used broad band seismograms observed at the Kamitakara observatory at distances of about 30 km. The events had seismic moments between 10**14 Nm to 10**17 Nm. To determine the source durations, we systematically searched for the rise times of the source time functions which minimized the difference between synthetic waveforms and observed data. In the data we could clearly see the near-field contributions to the waveforms, so it seemed that the durations were longer than expected for typical events, considering their size and the distance to the station. We estimated the source durations (rise time of source time functions) and source dimensions to study the source parameters of these event. For calculating synthetic waveforms, we used focal mechanisms determined by the FREESIA project and we assumed source time functions of symmetric triangles. Source dimensions are determined from these source durations under the assumption of circular faults and seismic moments are estimated from maximum amplitudes. These results show that source durations are longer and source dimensions are larger compared to most earthquakes with the same seismic moments.

2000 Japan Earth and Planetary Sciences Joint Meeting
Tokyo, June 25-28, 2000

Observations on the Fault Dynamic Rupture of the 1999 Chi-Chi (Mw7.7), Taiwan Earthquake
James Mori, K-F., Ma

Estimating Earthquake Source Dimensions in the Hida Mountains Region Swarm using Near-Field Waveforms
Sato, K. and J. Mori