AGU Fall Meeting
San Francisco, California, Dec. 11-15, 2006
Observations and Mechanisms for Continuous Tremor and Low-Frequency Events in Subduction Zones
The recent observations in Japan and Cascadia of continuous tremor and low-frequency events associated with subduction zones at depths of about 30 to 40 km, has sparked interest in processes that are occurring near the slab in this region. There is still little known about the physical mechanisms for these events, and even their position relative to the subducting slab is uncertain, because the low amplitude and emergent waveforms hamper precise locations. Some of the notable characteristics of the events include, episodic sequences, triggered occurrences from regional and teleseismic earthquakes, and propagation of events over distances greater than a hundred kilometers. Sometimes there are close associations of the occurrence of the events with slow slip events that are on the plate interface. Since it is difficult to analyze the waveforms of continuous tremor, much of the information is inferred from the discreet low-frequency seismic events, which are assumed to have similar mechanisms to the continuous tremor. From locations of low-frequency events, their position relative to the slab varies depending on the region. In some places low-frequency events occur close to the slab interface while in other places they are 10 to 20 km above the interface. Although there is little information about the focal mechanisms, since it is difficult to obtain moment tensor solutions for the small earthquakes, these events are often associated with fluid flow or pressure changes. This is inferred from the triggering stresses of teleseisms and the propagation across long distances. The long durations of continuous tremor that last for minutes to hours may be due to a fluid pressure change with feed back a mechanism that can continually induce further pressure changes.
Rupture Propagation of the July 17, 2006 Java Earthquake from Back Projection of Hi-Net Data
Mori, J. and S. Park
We used 717 short-period vertical components of the Hi-Net array in Japan (National Institute for Earth Science and Disaster Prevention) for a back projection analysis of the July 17, 2006 West Java earthquake. This method was previously used to show the rupture propagation along the long fault of the 2004 Sumatra- Andaman earthquake by Ishii et al. (2005), and in this study we have been able to obtain higher resolution to examine a smaller 200 km fault. The earthquake had a long duration of over 150 sec, so that there was sufficient time resolution to see the rupture propagation. The array is located at distances of 52 to 70 degrees from the earthquake and clearly records the direct P wave. Data were high-passed filter at 0.2 hz and aligned on the first arrival using waveform cross correlations. For the back-projection calculation, we chose a grid of 240 points in the source area and estimated which grid points were the sources of seismic radiation for sequential time windows of the P wave. The initial arrival of the first time window was assumed to come from the grid point corresponding to the earthquake hypocenter. For subsequent time windows, the data were stacked assuming a source at each grid point. The stack with the highest correlation was interpreted to be the source of the seismic radiation for that time window. We tried several values from 10 to 30 sec for the time windows and found the results to be fairly stable. Our results show an overall low rupture speed of about 1 km/sec for the earthquake, but the progression is irregular with areas of faster propagation. This suggests that the overall low rupture speed may be due to delayed multiple events and not a continuously slow rupture.
Quantifying Early Aftershock Activity of the 2004 Mid Niigata Prefecture Earthquake (Mw6.6)
Enescu, B., J. Mori, J., M. Miyazawa
We analyse the early aftershock activity of the 2004 Mid Niigata earthquake, using both earthquake catalog data and continuous waveform recordings. The frequency-magnitude distribution analysis of the Japan Meteorological Agency (JMA) catalog shows that the magnitude of completeness of the aftershocks changes from values around 5.0, immediately after the mainshock, to about 1.8, twelve hours later. Such a large incompleteness of early events can bias significantly the estimation of aftershock rates. To better determine the temporal pattern of aftershocks in the first minutes after the Niigata earthquake, we analyse the continuous seismograms recorded at six Hi-Net (High Sensitivity Seismograph Network) stations located close to the aftershock distribution. Clear aftershocks can be seen from about 35 sec. after the mainshock. We use events which are both identified on the filtered waveforms and are listed in the JMA catalogue, to calibrate an amplitude-magnitude relation. We estimate that the events picked on the waveforms recorded at two seismic stations (NGOH and YNTH), situated on opposite sides of the aftershock distribution, are complete above a threshold magnitude of 3.4. The c-value determined by taking these events into account is about 0.003 days (4.3 min). Statistical tests demonstrate that a small, but non-zero, c-value is a reliable result. We also analyse the decay with time of the moment release rates of the aftershocks in the JMA catalog, since these rates should be much less influenced by the missing small events. The moment rates follow a power-law time dependence from a few minutes to months after the mainshock. We finally show that the rate-and-state dependent friction law or stress corrosion could explain well our findings.
Low-frequency Earthquakes and 3-D Velocity Structure in the Bungo Channel and Shikoku Area, Japan
Nugraha, A. D., and J. Mori
We used a tomographic inversion to determine P and S wave structures
in the western Shikoku and Bungo Channel region, for
comparison with locations of low-frequency earthquakes. The velocity model
clearly images the high velocity subducting slab and
we can see the spatial relation to the low-frequency earthquakes. Under western
Shikoku the low-frequency earthquakes occur at depths close to the plate
interface, but under the adjacent Bungo channel
region to the west, there is a clear depth separation. Our interpretation is
that the low- frequency events are occurring in a region of high Vp/Vs that is located above the subducting
slab. We used local P and S phases to determine the three-dimensional velocity
structure for the region of Bungo channel and western
Shikoku. We used the program SIMULPS12 (Thurber and Eberhart-Phillips,
1993) which provide a local earthquake algorithm inversion to calculate the 3-D
Vp and Vp/Vs
structures. The iterative damped least-squares algorithm simultaneously
calculates the velocity model and hypocentral
adjustments. The data were P and S arrival times compiled by the Japan
Meteorological Agency (JMA) for 1998 to 2003. 2981 events with 655, 503 arrival
times from 208 stations in the area were used for inversion. Travel time data
of the low-frequency earthquakes (LFE) from the JMA catalogue were used for
July 2000 to December 2003 to relocate the events. Because of relatively large
uncertainties in the arrival time of the LFE, these data were not included in
the inversion, but were relocated using the 3-D structure determined from the
ordinary earthquakes. Our tomographic inversion for
the P and S wave structure for the subduction zone in
the western Shikoku and the Bungo channel clearly
identify the subducting slab from local velocity
increases of about 10%. The associated seismicity of ordinary earthquakes
appears to be located within the slab. In contrast, the depths of the
low-frequency events are 5 to 20 km shallower than the ordinary earthquakes
with the difference increasing toward the west. In the Bungo
channel area, the LFE events are located significantly shallower than the
imaged slab and we conclude that these events are occurring in a region of high
Vp/Vs that is located above the subducting
Change in Seismic Attenuation of the Nojima Fault Zone Measured Using Spectral Ratios from Borehole Seismometers
Kano, Y., K. Tadokoro, K. Nishigami, K., J. Mori
We measured the seismic attenuation of the rock mass surrounding the Nojima fault, Japan, by estimating the P-wave quality factor, Qp, using spectral ratios derived from a multi-depth (800 m and 1800 m) seismometer array. We detected an increase of Qp in 2003-2006 compared to 1999-2000. Following the 1995 Kobe earthquake, the project eFault Zone Probef drilled three boreholes to depths of 500 m, 800 m, 1800 m, in Toshima, along the southern part of the Nojima fault. The 1800-m borehole was reported to reach the fault surface. One seismometer (TOS1) was installed at the bottom of the 800-m borehole in 1996 and another (TOS2) at the bottom of 1800-m borehole in 1997. The sampling rate of the seismometers is 100 Hz. The slope of the spectral ratios for the two stations plotted on a linear-log plot is -Ξ t*, where t* is the travel time divided by the Qp for the path difference between the stations. For the estimation of Qp, we used events recorded by both TOS1 and TOS2 for periods of 1999-2000 and 2003-2006. To improve the signal-to-noise ratio of the spectral ratios, we first calculated spectra ratios between TOS1 and TOS2 for each event and averaged the values over the earthquakes for each period. We used the events that occurred within 10 km from TOS2, and the numbers of events are 74 for 1999-2000 and 105 for 2003-2006. Magnitudes of the events range from M0.5 to M3.1. The average value of Qp for 1999-2000 increased significantly compared to 2003-2006. The attenuation of rock mass surrounding the fault in 2003-2006 is smaller than that in 1999-2000, which suggests that the fault zone became stiffer after the earthquake. At the Nojima fault, permeability measured by repeated pumping tests decreased with time from the Kobe earthquake, infering the closure of cracks and a fault healing process occurred The increase of Qp is another piece of evidence for the the healing process of the Nojima fault zone.
Implications for Dynamic Rupture from Slow Slip Events
Brodsky, E. E. and J. Mori
The ratio of slip to fault length is more or less consistent for most earthquakes, yet the physical controls on this ratio are unknown. Recently discovered slow slip and creep events in which faults move quasi-staticly over hours to even years can shed new light on this old conundrum. These events have proportionately smaller amounts of slip compared to ordinary earthquake with similar fault dimensions. Eight out of the nine well- recorded creep events have stress drops <0.2 MPa while the mean stress drop for a set of ordinary earthquakes is 2 MPa. In terms of slip, we observe large slow events often have rupture dimensions of ~100 km and slip several tens of centimeters, while ordinary earthquakes that rupture a comparable length of fault typically slip meters. Tsunami earthquakes are intermediate between these extremes. The small slips of quasi-static events compared to the large slips in fast ruptures of earthquakes suggest that dynamic processes largely control the rupture growth of ordinary earthquakes. In particular the inertia of the dynamic slip is important in overcoming local barriers as the rupture extends. Dynamic models suggest that transient stresses on a rapidly slipping fault can overshoot the final stress drop by as much as 35%, so this difference between the dynamic and static stresses by itself is not enough to explain the observation. Another feature of dynamic cracks is that the rupture velocity is proportional to the square of the imposed stress difference. Again, the data support a much stronger dependence with rupture velocity varying at least as stress drop cubed. Therefore we conclude that a simple dynamic crack model in a homogeneous medium fails to explain the observation. However, on a fault with varying strength due to either a rough geometry or variable frictional processes, the dynamic stresses during earthquake rupture can allow slip to continue even when it encounters a high strength barrier. Slow slip events have no such advantage and therefore are easily stopped on heterogeneous faults. Therefore the average static stress drop for large ordinary earthquakes represents the difference between high strength barriers and the low dynamic friction, while the average static stress drop for slow slip events represents only the local stress minima on the fault.
Stress Drop and Radiated Seismic Energy of Microearthquakes Involving Volume Change in a South African Gold Mine
Yamada, T., R.E. Abercrombie, S. Ide, J. Mori, H. Kawakata, M. Nakatani, Y. Iio, H. Ogasawara
Source parameters of small earthquakes are important for clarifying the scaling relations between small and large earthquakes. However, it is often difficult to accurately determine source parameters of small earthquakes because the dominantly high-frequency seismic waves excited by small earthquakes are easily scattered and attenuated along the path. To overcome this problem we analyze the high frequency recordings of earthquakes at very close distances in the Mponeng mine in South Africa. We estimate the stress drop and radiated seismic energy of 20 microearthquakes (0.0 < MW < 1.3) that occurred in the mine to investigate their rupture characteristics and scaling relationships to large earthquakes. We analyze seismograms of borehole accelerometers recorded with high sampling rate (15 kHz) within 200 m of the hypocenters at the depth of 2,650 m. At all stations, the waveform data have very high signal-to-noise ratio and no significant later phases are observed. Corner frequencies and quality factors of the anelastic attenuation Q are estimated from spectra of the velocity seismograms by assuming the omega-squared model of Boatwright . We also investigate moment tensors for double couple solutions and volumetric components from the waveform inversion. 10 out of the 20 earthquakes have large volume changes and they range 30 through 50% of the moments for the best-fit double couples. Static stress drops of the 20 earthquakes calculated from the model of Madariaga  range from 3.2 to 88 MPa. Values of the scaled energy (= ER/Mo; the ratio of the radiated energy ER to the seismic moment Mo) with corrections due to volume changes and radiation patterns range from 4.2~10-6 to 1.1~10-4. We find that both the static stress drop and the scaled energy of the analyzed earthquakes are comparable to values for larger earthquakes and no scale dependence is observed. Our results indicate that the dynamic rupture processes of these microearthquakes are similar to those of larger earthquakes, even though they include significant volume changes.
4th International Symposium
Kyoto, Japan Dec. 2-5, 2006
The July 17, 2006 West
Java Tsunami Earthquake
A tsunami earthquake (Mw 7.7) occurred south of Java on July 17, 2006. The event produced relatively low levels of high-frequency radiation and local felt reports indicated only weak shaking in Java. There was no ground motion damage from the earthquake but there was extensive damage and loss of life from the tsunami along 250 km of the south coasts of Western and Central Java. An inspection of the area a few days after the earthquake showed extensive damage to wooden and unreinforced masonry buildings that were located within several hundred meters of the coast. Since there was no tsunami warning system in place, efforts to escape the large waves depended on the reaction of people to the earthquake shaking, which was only weakly felt in the coastal areas. This experience emphasizes the need for adequate tsunami warning systems for regions around the Indian Ocean.
We studied the rupture process of the earthquake using 717 short-period vertical components of the Hi-Net array in Japan@(National Institute for@Earth Science and Disaster Prevention) in a back projection analysis. The earthquake had a long duration of over 150 sec, so that there was sufficient time resolution to see the rupture propagation. The array is located at distances of 52 to 70 degrees from the earthquake and clearly records the direct P wave. For the back-projection calculation, we chose a grid of 240 points in the source area and estimated which grid points were the sources of seismic radiation for sequential time windows of the P wave. The initial arrival of the first time window was assumed to come from the grid point corresponding to the earthquake hypocenter. For subsequent time windows, the data were stacked assuming a source at each grid point. The stack with the highest correlation was interpreted to be the source of the seismic radiation for that time window. Our results show an overall low rupture speed of about 1 km/sec for the earthquake, but the progression is irregular with areas of faster propagation. This suggests that the overall low rupture speed may be due to delayed multiple events and not a continuously slow rupture.
The Foreshock Sequence of the
2002 Eastern Tottori Earthquake
Hiura, H. and J. Mori
Radiation Efficiency of
Intermediate-Depth Earthquakes (Mj>4.0) in the
Pacific Slab Beneath Japan
Nishitsuji, Y. and J. Mori
Velocity Structure in
the Bungo Channel and Shikoku Area, Japan and Its
Relationship to Low-frequency Earthquakes
Nugraha, A.D. and J. Mori
Determination of Rupture Velocity
of Deep-Focus Earthquakes using Regional Array Data
Park, S. and J. Mori
8th Kyoto University International Symposium, Towards Harmonious Coexistence with Human and Ecological Community on this Planet
Bangkok, Thailand, Nov. 23-25, 2006
Studies of the Active Geosphere: Faulting, Seismic
Structure, and Tsunamis
The Kyoto University Active Geosphere Investigations for the 21st Century Centers of Excellence (KAGI21) is a research and educational program that integrates a wide range of fields in geophysics. The study of earthquakes is an important part of this program and includes field investigations, laboratory experiments, and theoretical modeling. A large part of the focus of the whole KAGI21 research effort is on the role of water and observations of temperature in geophysics, and this also applies to the earthquake studies. Traditionally temperature measurements and estimates of water have not been used widely in seismological studies, so these studies represent exciting new areas of research.
Temperature Measurements on the Chelungpu
Following the September 21, 1999 Chi-Chi earthquake (Mw7.6), the Taiwan Chelungpu-fault Drilling Project drilled two boreholes which penetrated the fault at depths of about 1100 m in the northern part of the rupture zone. During the earthquake, this portion of the fault had a large amount of slip, 5 to 6 meters. The boreholes provided the rare opportunity to make temperature measurements in a fault zone with large slip from a recent earthquake. These type of observations are important to measure the amount of frictional heat that occurred at the time of the earthquake. To carry out the high resolution measurements, we developed borehole tools that contained quartz and platiumn resistance thermometers. Using these instruments, precise temperature measurements were recently carried out in one of the boreholes from March to September 2005, six years following the earthquake. The results showed a small temperature increase (0.06 oC) across the fault, which is interpreted to be representative of the frictional heat generated when the two sides of the fault move relative to each other at the time of the earthquake. In order to improve the interpretation of the temperature signature, we modeled the temperature distribution across the fault, including water flow and the spatial distribution of thermal conductivity. The modeling results imply an upper bound of the estimated heat generated by the earthquake, and indicate a very low level of friction at the time of the earthquake. The estimated temperature and frictional heat produced during the earthquake provide information on the dynamic process of faulting. These results are used in understanding the physical mechanisms that are needed to cause the low friction on the fault that allow large earthquakes to occur.
3-D Structure of the Slab and
The recent discovery of an unusual type of earthquake that contains only low-frequency seismic waves and is associated with subduction zones, has sparked interest in the structure of the subducting slab and the associated processes. These low-frequency earthquakes have been inferred to indicate water movement that is rising from the slab. Because these types of earthquakes are small and difficult to observe, their exact location relative to the subducting slab and their mechanism is still not fully known. We used a tomographic inversion to determine the P and S wave structures in the western Shikoku and Bungo Channel regions of western Japan. The velocity model clearly images the high velocity subducting slab and we can see its spatial relation to the position of the low-frequency earthquakes, which were relocated using the new 3-D model. Under western Shikoku the low-frequency earthquakes occur at depths close to the plate interface, while under the adjacent Bungo Channel region to the west, there is a clear depth separation with the low-frequency events occuring 10 to 20 km above the plate interface. Our interpretation is that the low-frequency events are occurring in a region above the slab interface. This region is characterized by a high ratio of the P-wave velocity to S-wave velocity, which may indicate the presence of water. The depth separation between the slab and the position of the low-frequency earthquakes varies depending on the region and may reflect the difference in the way the water is rising from the slab.
Observations of the West Java Tsunami Earthquake of
July 17, 2006
A tsunami earthquake (Mw 7.7) occurred south of Java on July 17, 2006. The event produced relatively low levels of high-frequency radiation and local felt reports indicated only weak shaking in Java. The KAGI21 International Summer School was being in Bandung, Indonesia when this earthquake occurred, so we were able to go quickly to the coastal area where the large tsunami occurred. We found that there was no ground motion damage from the earthquake but there was extensive damage and loss of life (over 600 people) from the tsunami along 250 km of the south coasts of Western and Central Java. An inspection of the area showed extensive damage to wooden and unreinforced masonry buildings that were located within several hundred meters of the coast. Since there was no tsunami warning system in place, efforts to escape the large waves depended on the reaction of people to the earthquake shaking, which was only weakly felt or not felt at all in the coastal areas.
Seismological Society of Japan Fall Meeting
Nagoya, Oct. 31 - Nov. 2, 2006
Estimates of Rupture Propagation
for the July 17, 2006 West Java Tsunami Earthquake
Mori, J. and S. Park
used short-period vertical seismograms of the P wave, as recorded by Hi-NetiNIEDj, to carry out a
back-projection analysis of the rupture for the July 17, 2006 West Java
earthquake (Mw7.7). Of the total 743 stations, we chose data from 717
stations that had generally similar waveforms, in order to eliminate stations
that had strong site responses or instrumental problems. The array is located
at distances of 52o to 70o from the earthquake and clearly records the direct P
wave. Data were high-passed filter at 0.2 hz
and aligned on the first arrival using waveform cross correlations. For the
back-projection calculation, we chose a grid of 240 points in the source area
and estimated which grid points best fit the sources of seismic radiation for
sequential time windows of the P wave. The initial arrival of the first time
window was assumed to come from the grid point corresponding to the earthquake
hypocenter. For subsequent time windows, the data were stacked assuming a
source at each grid point. The stack with the highest correlation was
interpreted to be the source of the seismic radiation for that time window. We
tried several values from 10 to 30 sec for the time windows and found the
results to be fairly stable.
The figure shows the results for time windows of 30 s offset by 15 s. Shown in each time window are the resulting locations of the waveform stacks that had the highest amplitude, and therefore inferred be the source of the P wave radiation. The differences in the stack amplitudes across the grid are quite small and vary by only a few percent or less, and for this reason, the stack amplitudes have been normalized in each time window in order to be able to show the locations of the maxima. The results clearly show the southeastward movement of the rupture. The rupture does not appear to have a smooth continuous propagation, but has jumps at faster rupture speeds with an overall average speed of about 1 km/sec.
Velocity Structure in the Bungo Channel and Shikoku Area, Japn
and Its Relationship to Low-frequency Earthquakes
Nugraha, A.D. and J. Mori
Quantifying early aftershock
activity of the 2004 Mid Niigata Prefecture Earthquake (Mw6.6)
Enescu, B., J. Mori, M. Miyazawa
The Radiation Efficiency of
Intermediate-Depth Earthquakes (Mj>4.0) in the
Pacific Slab Beneath Japan
Nishitsuji, Y. and J. Mori
Determination of Rupture Velocity
of Deep-Focus Earthquakes using Regional Array Data
Park, S. and J. Mori
Foreshock Sequence of the 2002
Eastern Tottori Earthquake
Hiura, H. and J. Mori
Urban Areas SCEC-ERI Workshop
Oxnard, California, June 1-3, 2006
Estimates of the
Slip-Weakening Distance and Fracture Energy from Near-field Records of the 1999
Chi-Chi, Taiwan Earthquake
The seismic fracture energy, or amount of dissipative energy during the rupture is a current topic in understanding the source process of earthquakes. The amount of energy may be related to such effects as breaking of contact points on the fault, reducing the grain sizes of fault rock, or creating new cracks in the vicinity of the fault. Some recent results suggest that there is a scaling with earthquake size. The seismic fracture energy is also related to the dynamic frictional level on the fault and related to the amount of heat produced.
We determined the amount of seismic fracture energy for the 1999 Chi-Chi, Taiwan earthquake (Mw7.6) by estimating the slip-weakening distance using near-field records. These records are located very close (within a few kilometers) to the fault and can be considered to be representative of the slip velocity and displacement for the nearby shallow portions of the fault. We used the observed records to estimate the slip-weakening distance for these regions of the fault that had large amounts of slip from about 3 to 7 meters. A simple parameterization of the source time function was implemented to represent different values of the slip-weakening distance. Synthetic seismograms are calculated to model the records for 4 near-fault stations, in order to determine the best fitting slip-weakening distance.
Our results show that the slip weakening distances are about 0.6 to 3.0 meters and there seems to be a dependence on the total amount of fault slip. These values for the slip-weakening distance imply that the amount of seismic fracture energy is a significant portion of the total energy and about two thirds of the radiated seismic energy. This gives a radiation efficiency of 0.6. The northern part of the fault that had larger slip appears to have a proportionately larger slip-weakening distance and thus a lower seismic efficiency, compared to the southern region of the fault.
We also used the near-fault records to estimate the rupture speed for the earthquake. The arrivals of the near-field waveforms indicate rupture speeds of 1.7 to 2.2 km/sec which correspond to about 0.6 to 0.8 times the shear velocity. The amounts of the seismic fracture energy as a function of rupture speed are also consistent with the direct determinations of the seismic fracture energy.
Workshop on Fault Zone Drilling
Miyazaki, May 23-26, 2006
Fault Characteristics, Energy Estimates, and Earthquake Recurrence: What One Seismologist Wants from Fault Drilling (Keynote)
There are some significant differences between the way seismologists and geologists typically look at faulting from large earthquakes. Seismologists tend to analyze waveform data which can usually resolve structures on the order of tens of kilometers, possibly down to one km is there is very good near-field data. Also, the derived physically properties are averaged over this scale length. Although it is difficult to resolve small scale features, seismologic studies have the advantage of seeing all depths of the fault. In contrast, fault zone geologists can make direct observations of the fault zone structures and properties, which are quite different from the remote sensing techniques of seismologists. Geologists tend to look at structures on the scale of microns to meters when examining the cores obtained from drilling into faults. These observations are spot measurements and it is often difficult to assess how representative they are of the entire fault. Also, fault zone sample studies are usually limited to a few kilometers depth, which may barely be in the range of the seismogenic zone that produces the seismic waves. These differences mean that seismologists and geologists are often looking at quite different aspects of the earthquake process. The approaches are different, but can also be complementary.
The depth of drilling is an important point for fault zone investigations. For comparing physical properties obtained in cores with seismological results obtained from waveform analyses, it is necessary to reach depths of the seismogenic zone, that is the regions of fault slip that have significant stress change to produce seismic waves. The shallow regions of faults, where materials have low rigidity, likely slip sympathetically with the seismogenic zone but do not produce seismic waves. The depth of the beginning seismogenic zone depends on local geology and probably is a transition zone. Waveform inversion studies suggest that the seismogenic zone may be as shallow as 1 km for the Chi-Chi, Taiwan earthquakes (Ji et al., 2001). It would be interesting if geologic studies could distinguish between seismic and non-seismic slip of the fault.
One of the fundamental issues in understanding the physical mechanisms of earthquakes, is clarifying the level of friction on the fault. One way to estimate the frictional levels during the faulting is to measure the heat produced. Measurements of the heat flow associated with the San Andreas fault have long been discussed over the past decades (e.g. summarized in Scholz, section 3.4.4). A more direct estimate would be to measure the fault zone temperatures immediately after a large earthquake. There were informal discussions have to measure fault-zone temperature following large the 1992 Landers and 2001 Denali earthquakes, although measurements were not done. Currently, the only available data of fault-zone temperatures following an earthquake are for the 1999 Chi-Chi, Taiwan earthquake (Kano et al., 2006). This result infers a very low level of dynamic friction during the earthquake, however, the measurement was made 5 years after the earthquake, when the temperature signal was quite small, and, it is difficult to make clear interpretations. To obtain better estimates of the fault-zone heat, measurements more quickly (a few months) after a large earthquake are necessary.
Fault slip during large earthquakes is usually heterogeneous with regions of large and small slip. The areas of large slip are often termed easpertiesf. These areas of large slip dominate the energy radiation and may control the rupture process. There are also suggestions that the stress accumulation during the interseismic period is different on the asperities, compared to the rest of the fault. The actual physical process that control the stress accumulation and large slip of the asperities is currently an active topic of discussion in seismology. Yamanaka and Kikuchi (2004) suggest that asperities are persistent features that are characteristic of a fault zone. However, results from the two recent Parkfield earthquakes and also large subduction earthquakes along the New Britain trench (Park and Mori, 2007) indicate that asperity distributions can be different for repeated ruptures of the same zone.
Some related seismological issues that might be answered by detailed analyses of fault zone structures from drilling are,.
E Are repeating earthquakes on a fault occurring on the same exact fault plane?
E Are the locations of asperities (areas of large slip) the same in repeating earthquakes?@
E What physically is the cause of asperities ?@
E How does each earthquake contribute to making the fault zone?
E What is the fracture/healing process on the fault?
Macroscopic and Microscopic Observations
Kanamori and Heaton (2002) emphasize the importance for understanding and integrating small and large scale results for fault zone studies. The physical processes that occur on microscopic scales of the fault zone are reflected in the macroscopic observations made by seismologists. The table lists some of the different and complementary aspects of seismological and borehole studies of earthquakes and fault zones. For example, detailed analyses of near-field seismograms for large earthquakes can give estimates of both the radiated and non-radiated energy. The non-radiated energy is thought to be dissipated by processes such as fracture formation and heat production. These dissipated processes can be studies with direct measurements of temperature, crack distributions, and grain sizes obtained from borehole observations.
Onshore and Offshore Drilling
The logistics of drilling crustal faults on land and drilling faults offshore, such as large megathrust subduction zones, are quite different and are addressed by the differences in the ICDP and IODP programs. Crustal onshore faults tend to be shallower and more accessible to drilling investigations. For large offshore subduction zones, identifying and reaching the fault surfaces with drilling is a difficult endeavor. Drilling both faulting environments is important for understanding the physical structures of the fault zones. There are also opportunities for cooperative onshore/offshore drilling projects that can combine observations of the same fault zone.
Finally to address the issues mentioned above, I would recommend that a fault zone drilling program would include the following (roughly in order of difficulty).
E Design projects to combine microscopic and macroscopic observations
E Drill to sufficient depth to reach the seismogenic zone (region that produces seismic waves)
E Drill soon after a large earthquake to measure temperature changes caused by faulting
E Use deep boreholes for measurements of earthquakes and strain (borehole observatories)
E Sample a fault in the same place before and after an earthquake
Precise Temperature Measurements
and Earthquake Heat Associated with the 1999 Chi-Chi, Taiwan Earthquake
Kano, Y., J. Mori, R. Fujio, T. Yanagidani, S. Nakao, H. Ito. K.-F. Ma,
Japan Geoscience Union Meeting 2006
Makuhari, Chiba, May 14-18, 2006
Estimates of Seismic Fracture Energy for Large Earthquakes
Mori, J. and K.-F. Ma
The seismic fracture energy, or amount of dissipative energy during the rupture, is a recent topic in understanding the source process of earthquakes. The amount of energy may be related to such effects as breaking of contact points on the fault, reducing the grain sizes of fault rock, or creating new cracks in the vicinity of the fault. Recent results (e.g. Abercrombie and Rice, 2005) suggest that there is a scaling with earthquake size. The seismic fracture energy is also related to the dynamic frictional level on the fault and related to the amount of heat produced. We estimate the amount of seismic fracture energy from near-field seismograms of the 1999 Chi-Chi, Taiwan earthquake (Mw7.6). These records are located very close (within a few kilometers) to the fault and can be considered to be representative of the slip velocity and displacement for the nearby shallow portions of the fault. We use measurements on the observed records to directly estimate the seismic fracture energy for these regions of the fault that had large amounts of slip from about 2 to 8 meters. We also carried out forward calculations of synthetic seismograms to investigate the resolution of the parameter estimation. If the slip weakening distance is less than one meter, it is very difficult to resolve the seismic fracture energy. Our results show that amount of seismic fracture energy is a significant portion of the energy and about 10 to 30% of the radiated seismic energy. The slip weakening distances are about 80 to 200 cm and there does not seem to be a strong dependence on the total amount of fault slip. We also used the near-fault records to estimate the rupture speed for the earthquake. The arrivals of the near-field terms, indicate rupture speeds of 1.7 to 2.2 km/sec which correspond to about 0.6 to 0.8 times the shear velocity. The results of the seismic fracture energy and generally consistent with the estimates of the rupture speed. A few near-field records for other large earthquakes (2000 Denali and 1999 Izmit) were used to check the consistency of the source parameter estimates.
Precise Temperature Measurement in a Deep Borehole Drilled in the Chelunpug Fault, Taiwan
Kano, Y., J. Mori, R. Fujio, H. Ito, T. Yanagidani, O. Matsubayashi, S.Nakao
have made a precise temperature measurement to observe a heat signature that is
associated with frictional heat generated at the time of faulting for a large
earthquake. Following the September 21, 1999 Chi-Chi earthquake, the Taiwan Chelungpu-fault Drilling Project (TCDP) bored two holes
which penetrate the fault at depths of about 1100 m near the town of Dakeng in the northern part of the rupture zone. During the
earthquake, this area had large surface rupture, and a fault displacement of
about 8 m is estimated from seismic data. The boreholes provided the rare
opportunity to make temperature measurements in a fault zone with large slip
from a recent earthquake. The precise temperature measurements were carried out
in one of the boreholes from March to September 2005, 5.5-6 years following the
earthquake. The borehole is cased with steel pipe so that there is no water
flow between the borehole and surrounding rock, enabling much more stable
temperature measurements. In order to obtain a high resolution temperature
profile, we developed a borehole instrument (quartz thermometer) containing two
quartz oscillator thermometers, separated by 3 m. We also developed a
temperature measurement system (platinum thermometer) using 5 platinum
resistance temperature detectors. We installed both quartz thermometers and
platinum thermometers in the borehole at depths between 1090 - 1111 m and made
a long-term temperature measurement at 7 depth levels for 6 months. On
September 2005, the quartz thermometer was slowly lowered (about 1.0 m/minute)
and raised (about 0.4 m/minute) in the borehole between the depths of 900 and
1250 m, producing four independent temperature profiles across the fault zone.
The continuous recording of temperature at 10 s intervals produced 5 to 15
readings per meter. Temperature measurements in the borehole shows a small temperature
increase (0.06 K) across the fault even 6 years after the earthquake, which is
interpreted to be associated with the 1999 Chi-Chi, Taiwan earthquake. In order
to improve the interpretation of the temperature signature, we modeled the a temperature distribution across the fault, including
(1) spatial distribution of thermal conductivity and (2) water flow. The
different conductivity values caused by different rock types and water content
can cause fluctuations in the observed temperature profile. The water flow in
the rock also can modify the temperature distribution. These modeling provide
us an estimation of the upper bound of heat generated by earthquake, which
leads to a very low level of friction at the time of the earthquake.