2001 Fall AGU Fall Meeting
San Francisco, December 10-14, 2001


Slip Distribution of the 2001 West India Earthquake
Mori, J, T. Sato, H. Negishi


We used the orientation and size of the fault determined  from our aftershock results to carry out an inversion of  teleseismic data for the slip distribution of the 2001 West  India earthquake. Previous inversions for this earthquake have been done by but these solutions did not use the constraints on the fault geometry that are now available.  Choosing the correct fault plane from the two nodal  planes of the focal mechanism and limiting the mainshock  source area to the size of the aftershock region affects  the slip distribution. We used 19 teleseismic P waveforms which were well distributed in azimuth in a finite-fault  inversion on a grid of 80 subfaults. Since the observed  waveforms look similar at all azimuths, we decided that there was not much time resolution and used only one  time window. The results of the inversion for the various   rupture velocities tested, did not show significant differences. We show the results for a rupture velocity of 2.9 km per sec. The results of the inversion show that the largest area of slip is close to hypocenter. This asperity is about 10 km x 20 km with a maximum slip of about 10 meters. The area of large slip in the region of  the hypocenter corresponds closely to the area of most severe damage in the villages east of Bhuj. This area  probably experienced very strong shaking from the  rupture of the asperity. Bhuj, is located more than 30 km   west from the closest portion of the fault and probably  experience somewhat lower levels of ground motions, as seen in the intensity distribution. The character of the  slip distribution appears different from other shallow  earthquakes of equal size. The area of the fault is small  for a Mw7.7 event. Comparing the slip distribution of the  2001 West India earthquake to the similarly sized (Mw7.7) 1999 Taiwan earthquake. The Taiwan earthquake is spread out over a larger area and shows a  more complicated slip distribution. These difference can also be seen in the teleseismic waveforms. The India event has a more compact waveform with a large  amplitude pulse at the beginning. The smaller source size  means the 2001 West India earthquake had a higher  static stress drop. This implies that the nearfield ground  motions were higher, although with shorter durations,  compared to the 1999 Chichi Taiwan earthquake.


Aftershock Distribution of the 2001 Gujarat, India Earthquake (Mw 7.7) from Temporary Field Observations: Small and Deep Orientation of the Fault Plane
Negishi, H., J. Mori, T. Sato, R.P. Singh, S. Kumar, N. Hirata


A large shallow earthquake (Mw 7.7) occurred in the western part of India on January 26, 2001, causing great damage. When the earthquake occurred, geologists and seismologists expected to see evidence for significant amounts of ground rupture along a fault. The surface trace, however, was not found in the damaged area indicating that the fault rupture did not reach the surface. It is important to know the location and orientation of the fault since this earthquake produced heavy damage in this area. We planned and carried out seismic observations in the field to determine the detailed distribution of aftershocks and the geometry of the fault plane. We installed 8 portable seismographs in the damaged area and aftershock recording was conducted during February 28 through March 28. The seismic array extends about 70 km in the north-south direction and 40  km in the east-west direction. Three-component velocity sensors with natural frequency of 1 or 2 Hz and 20 bit  digital recorders with 100 Hz sampling were deployed at  all eight stations. We recorded over 1,400 events during  the instruments deployment. P and S arrival times were  picked manually within a precision of about 0.02 and 0.1  seconds, respectively. Hypocenters were determined using the Joint Hypocenter Determination (JHD) program which simultaneously calculates hypocenters, station corrections and a one-dimensional velocity structure  (Engdahl et al., 1982). The initial velocity structure model, which is used by the National Geophysical  Research Institute (B.K. Rastogi, pers. Comm.), has been   used in the present study. The area of aftershocks, which may be interpreted as the area of the fault that ruptured  during the mainshock, has dimensions of about 40 km x 40  km, extending from about 23.3N to 23.7N in the north-south direction and 70.1E to 70.5E in the east-west direction. The distribution of the aftershocks  shows southward dipping at an angle of about 40  degrees, and the depth range about 10 to 35 km. This is interpreted to be the fault plane of the mainshock. The  size of the aftershock distribution of this event is smaller than that of 1999 Chichi Taiwan, which has the same  moment magnitude. This indicates that the size of the fault plane of the Gujarat event is relatively small and the  earthquake has a high static stress drop.



Estimate of the Absolute Value of Regional Stress prior to the 2000 Western Tottori Earthquake from Rotation of Focal Mechanisms
Yoshikawa, K, J. Mori, H. Katao


 On October 6, 2000, an earthquake (Mw=6.6) occurred in western Tottori Prefecture, Japan in an area where no active fault had been previously identified. There was no displacement on the surface but seismic observations showed that the event was a left-lateral strike-slip fault on a northwest striking plane with a length of about 20 kilometers and slip of about 1.6 meters. A temporary  short-period network of over 50 seismographs was  installed soon after the mainshock in a cooperative effort  by universities in Japan. The dense spacing of stations in  the aftershock region enabled us to determine  well-constrained P-wave first motion focal mechanisms.  The results of determining the focal mechanisms showed  a bimodal distribution of the P-axes in two azimuths. One  of azimuths was in the east-west direction. This is similar  to the P-axes direction for the mainshock and previous  seismicity and is interpreted to represent the regional  stress direction prior to the earthquake. The second azimuth has a more northwest direction. We interpret  that the rotation of the P-axes for these aftershocks is caused by the change in stress field due to fault slip during the mainshock. We estimated the level of absolute of stress prior to Tottori earthquake by using a fault model for the earthquake and calculating the amount of stress change at the locations that the rotation in P-axes  were observed. Comparing the calculated change in stress due to the earthquake with the rotation of the  P-axis gives us an estimate of the level of absolute stress  prior to the mainshock. The focal mechanisms of aftershocks showed a rotation of the P-axes of about 15 to 25 degrees. This relatively large change indicates that the level of stress prior to the earthquake was rather low, on the order of a few tens of MPa.



Difference in Rupture Process Between Shallow and Deep Earthquakes Estimated From Radiated Energy of Small Events
Kim, A, J. Mori


We estimated the seismic energy of small earthquakes using short-period data in Japan to understand thedifference of the source mechanisms and rupture process between shallow (0-50 km) and deeper (50-200 km) events. In this study we analyzed smaller events under Mw4.0 because it is thought that large shallow earthquakes may be more similar to deep earthquakes in  terms of possible thermal effects on the fault. We also estimated seismic efficiency and frictional stress of these  events to understand the rupture process. Although  there are some hypotheses about the mechanism of    intermediate to deep earthquakes, it is not well  understood how these earthquakes occur. To investigate   this issue we compare radiated seismic energy of shallow   with deep events. We used the integrated short-period  acceleration seismograms in the frequency range from   0.5-20Hz. Because the radiated energy of P wave is less   than 4 percent of total energy we calculated only S   waves energy from the average of the vertical, radial and   transverse components. To correct for the effects of  attenuation we calculated a depth and frequency  dependent attenuation relation. We also estimated the  site amplification factors for the stations using as a   reference those sites that had average S wave velocities  of 500m/s or grater in the upper 30m. Using the results obtained from above process we investigated scaling of   radiated seismic energy versus the depth and then  estimated the seismic efficiency to discuss the difference   in rupture process between shallow and deep events. The   radiated energies of deeper events tend to be larger than   for shallower earthquakes which may be an indication of the difference in stress regime or rupture properties.


Earthquake Initiations of the 2000 Izu Islands Earthquake Swarm
Sato, K and J. Mori


In recent years, earthquake initiation phases have been  often studied. Iio [1992,1995] and Ellsworth and Beroza   [1995] identified "slow initial phase" and "nucleation   phase", respectively, and concluded that it is part to the   earthquake source process. Oppositely, Mori and   Kanamori [1996] concluded it is an effect of anelastic attenuation and rupture is propagated rapidly from the  hypocenter. Sato and Kanamori [1999] introduced a  model which can explain these two observational results   using the Griffith fracture criterion. This model employed a   new parameter, "trigger factor", which controls the  rupture propagation velocity. While a small trigger factor causes a slow propagation, a large one causes an abrupt   propagation. Recently there have been studies to analyse earthquake initiations of micro-earthquakes using this model (Hiramatsu et al. [2000], Kato et al. [2001]). In this study, we analyse moderate earthquakes using the model and discuss source parameters of earthquake  initiations, such as the trigger factor and size of the initial  crack, and relationships between these parameters and   earthquake sizes. During the summer of 2000, an  earthquake swarm occurred in the Izu Islands region,  Japan, about 100 Km south of Tokyo. We installed several strong-motion seismometers in the region and recorded   moderate earthquakes near the sources. We use these    data to analyse earthquake initiations using the Sato and   Kanamori model. Because of the trade-off between  anelastic attenuation and these source parameters, we  need to make care estimates of the attenuation effects.  In this study, we estimate a whole-path average Q value   between the source region and the recording station  (KZA) before estimating initiation parameters using smaller   earthquakes. Unfortunately, the data recorded at the  KZA station did not trigger on many smaller events, so we   used data from a nearby station (KZB) for determining the   Q. We then estimate the relative Q difference between   KZA and KZB. By determining the effects of attenuation  separately from the initiation parameters, we can obtain reliable results about earthquake initiations.


Seismicity and Forecasting of the 1991Eruption of Mount Pinatubo: A Ten-Year Retrospective
Power, J A, R.A. White, T.L. Murray, E. Laguerta, E.G. Ramos, J. mori, B.C. Bautista, A.B. Lockhart


The 1991 eruptions of Mount Pinatubo provided a unique opportunity to observe seismicity and magmatic  processes associated with a very large explosive  eruption. The rapid deployment of a 7-station seismic   network around the volcano in April and May of 1991 and  deployment of a second network in early July after many  of the initial stations were destroyed by the eruptions on  June 13 - 15 were critical to developing these  high-quality observations. Data from these networks were  recorded and archived using a portable digital acquisition  and analysis system. The interpretation of the processes  associated with 1991 Pinatubo seismicity is enhanced in   many cases by the large size of the eruptions and associated signals, the well-documented magma mixing  event that triggered the eruptions, and the associated   field studies of deposits. The precursory seismic sequence  included a persistent cluster of Volcano-Tectonic (VT)  earthquakes 5 to 10 km NW of the vent, deep (z$\sim$30   km) Long-Period (LP) events in May and June, a  prominent swarm of shallow hybrid events associated with  magma ascent and emplacement of a lava dome on June 3 - 12, and a rapid increase in the size and occurrence of  LP events and volcanic tremor following dome  emplacement. Strong swarms of LP events and tremor often preceded and accompanied the plinian eruptions  from June 12 to 15, although only one seismic station  remained operative during this period. Post eruption  seismicity was characterized by widespread VT   hypocenters that surround the volcano and may define the magma chamber that fed the eruptions. In late June,  as continuous eruptive activity and VTs declined, tremor and LP events resumed in regular 7 to 10 hour episodes  that accompanied the production of large ash plumes.  Successful forecasts of the 1991 eruptions were based   largely on an observed shift in the locus of earthquake  hypocenters, increased seismic energy release, a shift  from VT to LP seismicity, visual observations of the  volcano, and measurements of S02 flux. In retrospect,  many of the patterns and processes revealed in the Pinatubo seismicity are frequently observed at restless  volcanoes and often form the basis for eruption forecasts  and hazard mitigation.


2001 Japan Seismological Society Fall Meeting
Kagoshima, October 24-26, 2001


Searching for Physical Mechanisms to Explain the Large Asperity of the 1999 Chichi, Taiwan Earthquake
Mori, J., H. Ito, M. Ando, K.F. Ma, M. Zoback

The 1999 Chichi, Taiwan earthquake (Mw 7.7) was the best instrumentally-recorded large earthquake in the world and provides extensive new data for looking at the physical mechanisms of the rupture process. A prominent feature of the earthquake was the area of large (~ 10 meters) and shallow (surface displacements of 7-8 meters) slip on the northern part of the fault. The level of high-frequency ground acceleration from this asperity was relatively low, considering the large displacements. The relatively low level of high-frequency radiation along with the large slip velocities and large displacements, suggests a low level of friction on the fault during rupture. Various physical mechanisms, such as fault melting, fault lubrication, or thermal pressurization have been proposed to explain the slip-weakening process that likely occurred on this portion of the fault.   The shallow location of the main asperity of the Chichi earthquake provides a rare opportunity to examine a fault on which a large amount of slip has recently occurred. We hope to drill a deep borehole (~3 km) into the fault, primarily to examine the physical properties of the fault surface. Close analyses of the fault properties will likely provide information about the physical mechanisms associated with the large amount of slip that occurred on this portion of the fault. Shallow boreholes (200-300m) have already been drilled into the northern and southern portions of the fault. Preliminary analyses of these cores indicates that physical properties of the fautl may control differences in the rupture dynamics of the earthquake. Determination of the frictional levels and slip mechanisms is important for understanding of how large earthquakes occur.



Slip Distribution of the 2001 West India Earthquake
Mori, J., T. Sato, H. Negishi


We carried out a source inversion for the slip distribution of the 2001 India earthquake (Mw 7.7) using teleseimic data and constrained the size and orientation of the fault plane from the distribution of locally recorded aftershocks. The results of the inversion show that the source area of the earthquake was relatively small and the slip distribution rather simple. The smaller fault area implies a high static stress drop for the earthquake. If the features of the teleseismic waveforms can be extrapolated to the nearfield, the strong-ground motions may have had very strong ground velocities at periods of a few seconds.

Aftershock Distribution of the 2001 Gujerat, India Earthquake (Mw7.7) from Temporary Field Obserations
Negishi, H., J. Mori, T. Sato, R.P` Singh, S. Kumar


Depth of Seismogenic Layer in Kachchh District, Gujerat State, India
Sato, T., J. Mori, H. Negishi

Absolute Value of Stress prior to the 2000 Tottori Earthquake
Yoshikawa, K., J. Mori, H. Katao



Difference in Rupture Process between Shallow and Deep Earthquakes Estimated from Radiated Energy of Small Events
Kim, A. and J. Mori

Earthquake Initiation Process: The 2000 Izu Islands Earthquake Swarm
Sato, K. and J. Mori


2001 Joint Geoscience Assembly (JGA)

International Symposium on Earthquake and Active Tectonics, International Symposium on East Asian Tectonics ( iSEAT)
Taipei, Taiwan, September 25-26, 2001

Initiation of the Chichi Taiwan Earthquake
Mori, J.

I examined the initial few seconds for the beginning of the September 21, 1999 Chichi, Taiwan earthquake (Mw7.7) on strong motion records at distances of 5 to 20 km. From the initial portion of the waveforms I make estimates of the rupture direction, slip velocity and, dynamic stress drop for the beginning of the earthquake. From  previous work, it is known that during the later times of the rupture, there are high values of slip velocity and dynamic stress drop on the northern part of the fault. These values are compared to the values that are obtained at the beginning of the rupture to see how the dynamic stress drop changes with time.

From the slip distribution models and the pattern of damage, it is apparent that the earthquake rupture has different characteristics with time.  During the first few seconds there is a moderate amount of slip on the fault and high levels of high-frequency radiation. Later in the rupture, there are much larger values of slip and slip velocity, but with proportionately lower levels of high-frequency radiation. In this study, these differences are quantifed in terms of the slip velocity and dynamic stress drop.

The initiation of the Chichi earthquake is also investigated using the model of Sato and Kanamori (1999). Using this model, I try to estimate the initial size of the crack which begins the earthquake.  With reasonable values of attenuation, the size of the initiation is estimated to have a radius of less than 50 m.

Seismological Results from the Chi-chi, Taiwan Earthquake: The Best Instrumentally Recorded Earthquake in the World

 Mori, J.  and K-F Ma


The September 21, 1999 Chichi, Taiwan earthquake (Mw=7.6) produced the best set of local and regional strong-motion records from a large earthquake. This shallow thrust earthquake 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, as well as, 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 x 1027 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 displacements (8-10m) were measured on the fault.

 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 gsmoothh producing low levels of high-frequency radiation. In contrast, the southern portion of the fault had much smaller displacements, but the level of accelerations were higher.  The damage to small 1 to 3 storey buildings was also much more severe in the southern region.

 The region of very large and very fast fault slip provides an important opportunity for studying the faulting mechanism in large earthquakes. The high slip velocity implies high dynamic stress drop. High dynamic stress drop means that either the driving tectonic stress is high, or the frictional stress is low. The hypothesis of low friction provides a good explanation for the esmoothf slip observed
There are several possible explanations for producing smooth (low friction) stress on the fault. These include, material properties of the fault, fault melting, fluid pressurization, and lubrication mechanisms. It is difficult to distinguish these explanation from the seismic data. The best way to study the problem is to obtain samples of the fault and examine its physical properties. This is the main reason we advocate drilling the Chelungpu fault. The Chichi earthquake produced large fault displacements at relatively shallow depths, which are well within the reach of a moderate drilling project. Examination of fault samples from the region of large slip should be able to distinguish between the various models for explaining how large slip can occur on faults. This would answer one of the fundamental questions that has persisted in seismology over the last 3 decades.


2001 Japan Earth and Planetary Sciences Joint Meeting
Tokyo, June 4-8, 2001



Scaling of Radiated Energy for Moderate Earthquakes in Japan.
Kobayashi, H., J. Mori


We analyzed 115 moderate (M4-6) earthquakes that occurred in the Japan region from March 1997 through October 2000. We used acceerations records of events from the K-Net system (operated by the National Research Institute for Earthquake Science and Disaster Prevention) which provide a good frequency bandwidth for estimating the radiated energy. The acceleration records were filtered and integrated to velocity.  Following Kanamori et al, (1993), the integrated value of the square of the ground velocity was used for the estimate of radiated energy.

We found that the local site response has a large effect on the radiated energy estimate. To correct for this problem, we estimated station corrections for all the sites used in the study.  To calculate the station correction, the sites were classified into gstiffh and  gsofth sites using the average shear-wave velocity in the upper 30 meters. Stiff sites were defined as having an average shear-wave velocity greater or equal to 500 m/sec. Radiated energy estimates were calculated for only the stiff sites, then individual site corrections were determined relative to this value.  Using the station corrections, the overall values of  the radiated energies were about a factor of 3 lower, than if no stations corrections are applied.  Also, the range of uncertainty for the energy estimate is considerably smaller when including the station corrections.

The results of the radiated energy of our study show that for shallow earthquakes there is a slight increase in the ratio of radiated energy to moment, as a function of earthquake size. The larger earthquakes appear to radiate proportionately more energy. This is similar to the observations of energy scaling for shallow earthquakes in California. Our results also show a depth dependence, with deeper earthquakes proportionately radiating more energy


Aftershock Observation in the Source Region of the 2001 Western India Earthquake
Negishi, H., J. Mori, T. Sato


Scaling of Radiated Energy for Moderate Earthquakes in Japan
Kim, A., and J. Mori