Seismology [S]

S41B MCC:level 2 Thursday 0800h

Earthquakes in Mines as a Natural Laboratory of Earthquake Processes Posters

Presiding:Z Reches, University of Oklahoma; T Jordan, University of Southern California

S41B-0962 INVITED 0800h

Modes of Rupture and Fault Maturation Processes in South African Gold Mines

* Richardson, E ( , Penn State University, 406 Deike Bldg, University Park, PA 16802 United States
Nyblade, A A ( , Penn State University, 406 Deike Bldg, University Park, PA 16802 United States

Deep mines are the best environment for studying small events (moment magnitude $-2 \le M \le 3.5$) whose dimensions are at or below the detection threshold of most surface-based seismic arrays. Mining-induced seismicity includes events directly triggered by blasting that are assumed to involve fresh fracturing of rock as well as those induced over longer timescales that have been hypothesized to be dominated by frictional slip. These distinctions have so far been based on spatio-temporal clustering statistics and spectral signatures of these two types of events. We have inverted for the moment tensors of 17 events from the Far West Rand gold-mining region of South Africa that range in size from $1.0 \le M \le 3.2$. These events occurred at 1-4 km depth and were recorded locally by four networks of 102 three-component geophones installed at depth throughout the active mining environment as well as regionally by a two-year Passcal deployment of 80 broadband seismometers. The moment tensors of these events are consistent with purely double-couple solutions. Therefore, we assert that these events are indeed proxies for natural tectonic earthquakes that nucleate via friction-dominated slip on planar surfaces. For each fault that produced one of these 17 events, we have constructed frequency-magnitude curves that span two to five years of seismicity on that particular fault. Since mine operations force seismic activity at a relatively fast rate, we can use this information to study the evolution of faults. Specifically, our study shows that many faults underground mature over time. They begin as individual patches that produce little seismicity or fresh-fracturing seismicity alone and gradually become large faults capable of producing large friction-dominated events. We hypothesize that this process is tied to the evolution of fault gouge and repeated slip events on these faults.

S41B-0963 INVITED 0800h

Induced Seismicity at the Canadian Underground Research Laboratory 1987-2004

* Young, R P ( , University of Toronto, Lassonde Institute, 170 College Street,, Toronto, M5S3ES Canada
Collins, D ( , University of Toronto, Lassonde Institute, 170 College Street,, Toronto, M5S3ES Canada

The Underground Research Laboratory (URL) is a facility established by Atomic Energy of Canada Ltd. (AECL) in Manitoba to investigate the rock mechanical and geotechnical aspects of the safe geological disposal of radioactive waste. It represents a unique facility to study the fundamental behaviour of initially unfractured granite in situ. The facility consists of a network of tunnels that have been excavated at a depth of 420 meters below the surface. As part of the extensive suite of instrumentation installed to monitor the rock mass behaviour both a microseismic (detecting events in the -4 $<$ M $<$ -1 range) and an ultrasonic (detecting events in the -7 $<$ M $<$ -5 range) system were installed. These systems will detect the formation of, and movement on, cracks ranging in size from cms down to the grain boundary scale. During the excavation of the tunnel network at the 420m level of the URL an extensive data-set of induced seismicity data was recorded from which an understanding of the rock mass response was obtained. This seismicity was concentrated in zones of `excavation disturbance' around the tunnels, and the activity level was observed to fall away to almost nothing a few days after the excavation was complete in each tunnel section. However these disturbed zones are observed to be only meta-stable, as further perturbations of the system result in a re-activation of the damage zone. This has been observed, for example, when; (1) Loose material has been removed from the tunnel floor resulting in observed seismicity in the rock beneath the floor; (2) Activity associated with heating of the rock during a series of "heated-failure tests"; (3) Activity in the roof and floor of nearby tunnels during the excavation of new galleries. The system can therefore be considered to be at best meta-stable, with very small changes in the stress resulting in induced seismic activity. This fact, together with the extensive instrumentation suite available with which to detect small changes in the rock response, makes this site ideal for studying remotely triggered seismicity. Results from studies since 1987 and the insights gained for earthquake seismology and future proposed underground earth science research facilities will be described.

S41B-0964 0800h

AE analysis in developing the Hot Fractured Rock geothermal power in Australia

* Aoyagi, Y ( , Central Research Institute of Electric Power Industry, 1646 Abiko, Abiko-shi, Chiba-ken, Abiko, 270-1194 Japan
Kaieda, H ( , Central Research Institute of Electric Power Industry, 1646 Abiko, Abiko-shi, Chiba-ken, Abiko, 270-1194 Japan
Asanuma, H ( , Tohoku University, 6-6-20 Aoba, Aramaki-aza, Aoba-ku, Sendai-shi, Miyagi-ken, Sendai, 980-8579 Japan
Wyborn, D ( , Geodynamics Co.Ltd., Level 2 349 Coronation Drive MILTON QLD 4064, Brisbane, 4064 Australia

The hot fractured rock (HFR) geothermal power is being developed in Cooper Basin, South Australia since 2002. HFR geothermal power is one of natural energy acquiring systems, in which water is pumped into hot, crystalline rock via an injection well, becomes superheated as it flows through open joints in the hot rock reservoir, and is returned through production wells. At the surface, the useful heat is extracted by conventional processes, and the same water is re-circulated to mine more heat. Such hot granites are buried beneath 3.7 km of insulating sedimentary rocks at the site. The temperature of the granites reaches 250_E#381; or more. The first injection well Habanero#1 was drilled 720m into the granite, and a reservoir was made by the hydraulic fracturing in the vicinity of the well bottom (4421m in depth) in 2003. During the hydraulic fracturing many acoustic emissions (AE) were generated. We observed the AE activity using seismic network deployed in 8 wells around Habanero#1 to evaluate the reservoir. Total of 12000 or more AE were observed during the fracturing period from November to December, 2003. Although the AE hypocenters were located in the south side of the well at the initial stage, they finally distributed N-S to NE-SW direction at about 3km in diameter. The magnitude of the AE ranges M-2 to M1 for most events, but several felt earthquakes as maximum size of M3.7 were also generated. The hypocenters of the larger 12 events (> M2.5) were located by the seismic network of Geoscience Australia. The mechanism solution of these large events is basically E-W compression type, and it almost agrees to the regional stress estimated by borehole breakout in wells in the area. The AE generation property will help to understand earthquake dynamics and mechanics since it is controlled by hydraulic pressure. We will mainly discuss the relation between the generated regional energy and the mechanism solution of the events.

S41B-0965 0800h

Apparent Stresses of Mining Earthquakes with Seismic Moments of $10^{8}<M_O<10^{11}$Nm.

* YOSHIMURA, M ( , Gradurate school of Natural science and Technology, Kakuma, Kanazawa, 920-1192
HIRAMATSU, Y ( , Gradurate school of Natural science and Technology, Kakuma, Kanazawa, 920-1192
FURUMOTO, M ( , Gradurate school of Natural science and Technology, Kakuma, Kanazawa, 920-1192

Mining at a deeper part in the crust usually causes micro- to small earthquakes in the close vicinity of stopes. Seismogenic processes of earthquakes can be monitored at very short hypocentral distances with sensors installed in seismogenic areas. However, it has been controversial if a mining earthquake and an ordinary earthquake can be treated as an identical earthquake phenomenon. Richardson and Jordan(2002) shows that mining earthquakes are classified into two kinds from their characteristics and that the two kinds are different in a scaling relation between the apparent stress and the seismic moment. The purpose of this report is to determine precise apparent stresses of many mining events using high quality data and to compare them with those of natural earthquakes. To observe mining events, nine three components borehole accelerometers were installed in Precambrian quartzite (typical Young's modulus of 70Gpa) within 200 m along a haulage tunnel at a depth of 2650m in Mponeng mine in South Africa. The observation was carried out from February to December in 1996. More than 25 thousand seismic events were recorded with a sampling frequency of 15 kHz and a dynamic range of 120 dB. We selected 378 events with high S/N from the events. The hypocentral distance of the events, which are determined by assuming an infinite medium with the P-wave velocity 5.5 km/s and the S-wave velocity 3.2 km/s, are in a range of 15m to 1km. We estimate the apparent stress $\Delta \sigma _a$ of the events by applying the same method that was used by Ide and Beroza(2001). The obtained apparent stresses of the mining events are $0.4< \Delta \sigma _a<8MPa$ for events with the seismic moments $10^{8}<Mo<10^{11}$Nm. The apparent stresses are identical to those of natural earthquakes and are independent of earthquake size. There is no difference between mining earthquakes and natural earthquakes.

S41B-0966 INVITED 0800h

Semi-controlled Earthquake-generation Experiments to Monitor the Entire Life Span of an Earthquake in South African Deep Gold Mines

* Ogasawara, H ( , Fac Sci Engr, Ritsumeikan U, Noji, Kusatsu, 525 Japan
Nakatani, M , U Tokyo, Tokyo, Japan, 113
Iio, Y , DPRI Kyoto U, Uji, Japan, 611
Ishii, H , Tono Res Inst Eq Sci, Mizunami, Japan, 509
Yamauchi, T , Nagoya U, Nagoya, Japan, 464
Takeuchi, J , Fac Sci Engr, Ritsumeikan U, Noji, Kusatsu, 525 Japan
Shimoda, N , Fac Sci Engr, Ritsumeikan U, Noji, Kusatsu, 525 Japan
Kawakata, H , DPRI Kyoto U, Uji, Japan, 611
Mendecki, A J , ISS International, Stellenbosch, South Africa, 7613
van Aswegen, G , ISS International, Stellenbosch, South Africa, 7613
Kato, A (IFREE) , U Tokyo, Tokyo, Japan, 113
Satoh, T , Geol.Survey, AIST, Japan, 305
Otsuki, K , Tohoku U, Sendai, Japan, 980
Yamada, T , DPRI Kyoto U, Uji, Japan, 611
Kita, S , Tohoku U, Sendai, Japan, 980
Kuwano, O , U Tokyo, Tokyo, Japan, 113
Nagata, K , U Tokyo, Tokyo, Japan, 113
Morishita, K , Fac Sci Engr, Ritsumeikan U, Noji, Kusatsu, 525 Japan
Ide, S , U Tokyo, Tokyo, Japan, 113
Kusunose, K , Geol.Survey, AIST, Japan, 305
for SeeSA, T , Semi-controlled Earthquake-generation Experiments, in South African, deep, gold mine

Mining takes place at depths of 2-3 km in South Africa, thereby inducing events with M \textgreater 3 in the close vicinity of stopes, with the largest events so far recorded being M \textgreater 5. As a result, seismogenic processes can be monitored at very short distances with sensors installed ahead of time in seismogenic areas. We refer to this process as a semi-controlled earthquake-generation experiment, which cannot be done with natural earthquakes monitored from the Earth's surface. In the 1970s-1980s, pioneering work (e.g. McGarr et al. 1975) yielded abundant, fundamentally important results in this area. In more recent times, broad-band and wide-dynamic-range monitoring has enabled us to study additional details of the seismogenic process. Therefore, we have attempted to monitor the entire life span of an earthquake within a hypocentral distance of a few hundred meters. To date, we have monitored in six experimental fields in South African mines, the pilot field being near a strong dike 1700-m deep in a mine. The second was a homogeneous area 2700-m deep without existing faults or dikes in another mine. From 2000 we began to continuously monitor normal and shear strains on faults with 25-Hz and 24-bit sampling, where an event with M about 3 is expected at 2400-2600 m deep. Then, we have successfully monitored the entire strain history \textgreater 1E-4 in a fault loss associated with a few seismic events with M \textgreater 2. However, there were no close strong motion meter available to locate asperities; only a single strainmeter was available, not enabling us to locate strain change source; no in-situ stress measurements were carried out at the site; no information available to constrain strength. In order to solve the problems, from 2003 to 2004, we deployed new experimental fields at fault bracket/stabilizing pillars in South African deep gold mines 2900-m deep. We installed multiple strainmeters, arrays of strong ground motion meters, sensitive thermometer array to monitor seismic heat generation, and fault displacement meters. At the field, dense thermometer array to monitor strength of a fault was successfully deployed. In this paper, we review our activities to date and future prospects.

S41B-0967 0800h

Near-Field High-Resolution Seismic, Strain and Displacement Measurements for Earthquake Source Studies in Deep Mines in South Africa

* Johnston, M J ( , U. S. Geological Survay, 345 Middlefield Rd., MS977, Menlo Park, CA 94025 United States
Reches, Z ( , University of Oklahoma, 731 Elm Avenue, Norman, OK 73019 Israel
van Aswegan, G ( , ISSI, P.O. Box 12063 , Die Boord, 2501 South Africa
McGarr, A ( , U. S. Geological Survay, 345 Middlefield Rd., MS977, Menlo Park, CA 94025 United States
Lockner, D ( , U. S. Geological Survay, 345 Middlefield Rd., MS977, Menlo Park, CA 94025 United States
Sellers, E ( , MinTek, PO Box 91230, Auckland Park, 2006 South Africa
Ben Zion, Y ( , USC, Dept. Earth Sciences, Los Angeles, CA 90089 United States
Williams, C ( , U. S. Geological Survay, 345 Middlefield Rd., MS977, Menlo Park, CA 94025 United States

Unique access to information on the physics of the earthquake source (earthquake nucleation, fault rupture, heat generation, stress state, seismic wave propagation, fault displacement, material properties and particularly changes in some of these parameters prior to rupture) exists in the near-field of mining-induced earthquakes in deep gold mines in South Africa. The new NSF funded Natural Earthquake Laboratory in South African Mines (NELSAM) will provide seismic velocity/acceleration, ground strain, temperature, fault displacement, acoustic emission (AE), and perhaps self-potential (SP) data in small 3-D arrays across and within active faults in two different mines. 3-component accelerometers to be installed in or near the faults have a range from micro g to 0.5 g in the band 0.05 - 500 Hz. Fault displacement meters (creepmeters) to be installed at low angles across faults within boreholes have a range of microns to 0.2 m and cover the frequency range from DC to 100 Hz. Successful measurement of total displacement will depend on the creepmeter reference length surviving the fault rupture. Temperature will be measured to millidegrees C at points within, and at increasing distances from fault zones, to capture the heat generated by future and past earthquakes. Strain transients will be measured with 3-component near-fault borehole strainmeters with capacitance displacement transducers providing a resolution < 10E-9. 3-component seismic velocity transducers will be installed together with each accelerometer and supplement the current mine seismic network. AE and SP will be measured within boreholes crossing faults if recording capability is sufficient. All data will be digitally sampled and transmitted to the surface in real-time for analysis to focus on unraveling the physics of the nucleation process, non-linear deformation prior to rupture, propagating aseismic slip, and variation in the material properties of near-fault materials (e.g. state/rate dependent friction). Similar high-resolution borehole strain and seismic networks on the San Andreas fault system in California can make measurements no closer than 5 km from moderate to large earthquakes on these faults. However, these San Andreas data do provide a reference set for comparison with pre-event strain levels, earthquake nucleation moments, coseismic strains, aseismic slip events ("slow" earthquakes), fault creep events, changes in fault zone material properties and the basic rupture nucleation process.

S41B-0968 0800h

The Pretorious fault, Mponeng mine, South Africa: A site selected for the establishment of a natural earthquake laboratory

* Reches, Z ( , School of Geology and Geophysics, University of Oklahoma, Norman, OK 73019 United States
van Aswegen, G ( , Integrated Seismic Systems International, Western Deep Levels, Carleton, 7613 South Africa
Jordan, T H ( , Southern California Earthquake Center, University of Southern California, Los Angeles, CA 90089-0742 United States
Johnston, M ( , U.S. Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025-3591 United States
Zoback, M ( , School of Earth Sciences, Stanford University, Stanford, CA 94305-2215 United States

One major obstacle in earthquake investigations is the lack of direct and near-field observations. To reduce this limitation, we initiated the NELSAM project (Natural Earthquake Laboratory in South African Mines) for in-situ seismic observations at 3.5 km depth (session S11 this meeting). The project includes instrumentation of the complete frequency range (creepmeters, seismometers, strain meters, accelerometers, temperature sensors, acoustic emission transducers). This dense 3D array is designed to monitor fault activity before, during, and after M = 2.5-3.5 earthquakes at distances of 1-100 m from their anticipated hypocenters. The site characterization will include 3D mapping, in-situ stress measurements on local and regional scales, and rock mechanics analysis. We describe here the structural and seismic features of the first site where monitoring is planned to start circa April, 2005.We screened and ranked 12 active faults in South African mines according to dimensions, internal structure, accessibility, and likelihood of M 3.0 seismic events. The selected site is on the Pretorius fault, one of the largest faults in the Western Deep Levels gold mines, WSW of Johannesburg. This fault-zone is at least 10 km long; it is very steep and inclined toward the SSE with throw up to 70 m. It contains many individual fault surfaces that vary in inclination from 40 deg to vertical; some have thin breccia zones with quartz cementation, and a few display a fine-grain gouge, but there is no major, well-developed gouge zone. The Pretorius fault accommodated some amount of strike-slip displacement as indicated by frequent inversion of the sense of fault throw, the occurrence of horizontal slickenside striations, and its steepness. The Pretorius fault-zone was active in Archaean times, and it has been locally reactivated by the mining operations in Mponeng and Tautona mines, which provide access to depths of 3.0-3.5 km. Our monitoring site will be established where both sides of the Pretorius fault will be mined during the next few years; the practice in the deep mines and numerical modeling predict profound increase of the seismic activity at the site during the next 2-4 years. The associated increase of shear stresses on the fault is expected to generate a few earthquakes of M>3.0 along surfaces of hundreds to thousands of sq. meters within the fault zone. Monitoring the fault zone activity and particularly the larger events is the main objective of NELSAM.

S41B-0969 0800h

Alteration of fresh fault gouge from focal depths of recent earthquakes in deep mines

* Dewers, T ( , School of Geology and Geophysics, University of Oklahoma, 100 East Boyd Street, Suite 810, Norman, OK 73019 United States
Reches, Z ( , School of Geology and Geophysics, University of Oklahoma, 100 East Boyd Street, Suite 810, Norman, OK 73019 United States

Rock powder recovered from focal depths in South African gold mines has been used as an estimate of the surface energy released during earthquakes (Olgaard and Brace, 1983). Recent work (Wilson et al., this meeting) has shown the amount of new surface area generated by earthquakes can approach 100 $m^2/g$, representing a sizable fraction of earthquake energy budgets. Chemical alteration of gouge, including weathering reactions, simple dissolution, precipitation of new phases, and Ostwald ripening, will act to decrease specific surface area and thus mask estimates of released surface energy from a single earthquake with time post-event. To assess this question of gouge textural alteration/preservation, we examine alteration rates of quartzose fault gouge sampled at 2.7 km depth from the Bosman Fault, newly-borne during the 1997 M=3.7 earthquake in Hartebeestfontein gold mine, South Africa, which accumulated ~0.4 m of total slip. TEM analysis of the rock powder suggests no presence of glass or alteration mineralogy. Particle size distributions of starting material yield a broad range with mean of about 0.1 microns. Hydrothermal alteration experiments were performed at 50, 75, 100 and 150 degrees C at 0.7 MPa fluid pressure in a continuously stirred titanium pressure vessel. Notable changes in experimental products include an increase in mean size of distributions ($>$ 1 micron for longest run times) after weeks to months depending on temperature and a shift to fewer, larger particles in the distribution, consistent with Ostwald ripening. Aqueous silica concentrations from timed experiments at the higher temperatures show an initial increase above quartz solubility followed by asymptotic decay to saturation levels. We apply a particulate dynamics approach to assess the kinetics of this process with a PSD-dependent mineral-water reaction rate law, parameterized by observed textural and solution chemical changes. Inasmuch as this type of particulate growth mechanism can be recognized by attributes of the particle size distribution (Eberl et al., 1998), this model is then used to assess amount and mechanism of alteration in gouges from fault zones in other South African gold mines as well as an exhumed cataclasite zone of the San Andreas Fault at Tejon Pass, California USA. In the NELSAM project (Johnston et al., this meeting, session S11), fresh gouge will be recovered from drill core within months of an earthquake event to monitor total specific surface area produced and energy released, and to test the validity of a rapid alteration hypothesis.

S41B-0970 0800h

Master Event Relative Location of Excavation Induced Micro Seismicity at the Underground Research Laboratory.

Reyes-Montes, J M ( , Department of Earth and Ocean Sciences, University of Liverpool.4 Brownlow Street, Liverpool, L69 3GP United Kingdom
* Rietbrock, A ( , Department of Earth and Ocean Sciences, University of Liverpool.4 Brownlow Street, Liverpool, L69 3GP United Kingdom
Collins, D S ( , Lassonde Institute, University of Toronto. 170 College Street, Toronto, M5S 3E3 Canada
Young, R P ( , Lassonde Institute, University of Toronto. 170 College Street, Toronto, M5S 3E3 Canada

Microseismicity has been extensively used to non-destructively monitor induced damage accumulation and stability of underground structures. Source location is one of the most fundamental characteristics of seismicity in order to provide this information. However there is an uncertainty in its determination, arising from a combination of three factors: (1) measurement error, (2) velocity model error, and (3) a non linear term resulting from the second and higher order terms in the Taylor polynomial expansion of the travel time equation. Classical location routines in this context cannot resolve events separated less than 0.5 m, in the order of magnitude of the largest fractures observed. Master event relative locations of the events recorded during the excavation of the TSX tunnel at the Underground Research Laboratory enhance this resolution up to one order of magnitude, as shown by the location of synthetic events performed in this context. This approach implicitly removes the uncertainty introduced by the anisotropic P wave velocity field induced by the presence of the tunnel and the damaged rock. This study presents the relocation of 1468 MS events recorded at the TSX during the excavation phase between February and March 1997. For the relocation, events are divided into two main clusters in each excavation section, according to their initial spatial location (performed using Geiger's routine and an average P wave velocity) corresponding to different regions in the tunnel perimeter, the tunnel floor and the roof. Master events are chosen among those with the maximum number of P wave arrival times available, sharpness of onsets, and being centrally located with respect to each cluster of events. A precision higher than 0.05 m was achieved on the location of synthetic events less than 1 m apart from the master in this context. The relocated events locate closer to the excavation perimeter than the absolute locations, a result that corresponds with the observed damage zone. Additionally the relocated events define a clearer structure within each cluster, with planes of alignment parallel to the tunnel perimeter. A series of hits performed in August 2003 using a Proceq Schmidt hammer source in the tunnel perimeter were used as masters for a subset of 323 events. This provides a repeatable, accurately located source of reference events. The relocated events show a similar trend towards shorter distances as in the previous set. Also the structure within the cluster is consistent with the one observed using the induced events as reference.

S41B-0971 0800h

Analysis of Regional Seismic Waveforms From Mining Explosions in the United States and Russia

* Renwald, M D ( , Southern Methodist University Department of Geological Sciences, POB 750395, Dallas, TX 75275-0395 United States
Arrowsmith, S J ( , University of California, San Diego Scripps Institution of Oceanography, 9500 Gilman Drive, La Jolla, CA 92093-0225 United States
Stump, B W ( , Southern Methodist University Department of Geological Sciences, POB 750395, Dallas, TX 75275-0395 United States
Hedlin, M ( , University of California, San Diego Scripps Institution of Oceanography, 9500 Gilman Drive, La Jolla, CA 92093-0225 United States

Mining explosions provide an opportunity to separate the source and regional path effects on observed seismic waveforms. We study mining explosions and their accompanying regional seismic waveforms in two separate regions, the continental United States and the Altai-Sayan region in Russia. The US is chosen as a study area because: (1) significant ground truth information already exists; (2) high-quality broadband data for both delay-fired and contained single-fired explosions are readily available; (3) multiple, high-performance regional arrays already exist or are planned; and (4) the US can be divided into two distinct regions of propagation path - high attenuation in the west and low attenuation in the east. We have assembled a preliminary database which takes advantage of the aforementioned characteristics and includes mining explosions from copper, coal, and iron mines in Arizona, Wyoming, and Minnesota, as well as nearby shallow low-magnitude earthquakes. The second region of study, the Altay-Sayan in Russia, is chosen because it is one of the largest and most active mining trends in the world. The Altay-Sayan is monitored both by single stations and arrays in the IMS and the GSN as well as the high-density regional Altay-Sayan Seismological Expedition (ASSE) network. The Russian portion of the database contains seismic waveforms from 500 known mine blasts in the Altai-Sayan using five stations in the IMS and GSN seismic networks: ZAL, MAKZ, KURK, BRVK and TLY and will be complimented with data from the ASSE seismic network in the near future. By utilizing these waveform databases, we aim to characterize the regional propagation path effects in both study areas and to link the recorded waveforms to the known physical source processes. Initial investigation of source effects manifested in waveforms shows large differences in regional amplitudes due to both source timing and explosion geometry.