The ANZA Broadband Seismic Network: 24-Bit Real-Time Telemetry Network

Contents

• Introduction
• Seismic Hazards of the Region
• Seismicity Recorded by the ANZA Network
• Network Instrumentation
• Streckeisen STS-2 Broadband Seismometers
• 24 Bit Digitizers and Real-Time Telemetry
• Toro Peak Relay Station
• Real-Time Processing
• Data Review and Archive
• Research Results Based on ANZA Data
• P and S Tomographic Velocity Structure Inversion
• Small Scale Stress Heterogeneity
• Temporal stability of travel times
• Nucleation processes of large earthquakes
• Coherence studies
• Project Plan
• Proposed Operations
• Additions to the ANZA network
• Data Processing
• Reports and Dissemination of Information and Data
• Related Efforts
• Project Personnel
• Institutional Qualifications
• ANZA Bibliography 1992-Present
• References

NOTE
No figures in this copy of the proposal as yet

The ANZA Seismic Network utilizes state-of-the-art broadband sensors and 24-bit dataloggers combined with real-time telemetry to monitor local and regional seismicity in southern California. The goal of this project is to provide on-scale digital recording of high-resolution three component seismic data for all earthquakes.

The ANZA network is centered around the Anza segment of the San Jacinto fault zone which has a maximum expected characteristic earthquake magnitude of 7.5. There is a high level of microseismicity (Ml < 4.5) in the Anzaregion. It is also located in a region where there is a large number of significant events. The 1986 North Palm Springs (Mw = 6.2), 1987 Superstition Hills (Mw = 6.5), 1987 Elmore Ranch (Mw = 5.9), 1992 Joshua Tree (Mw = 6.1), 1992 Landers (Mw = 7.3), and 1992 Big Bear (Mw = 6.2) have all occurred within 100 km of the center of the ANZA network since it was installed in 1982. In addition, the Southern California batholith is widely exposed on both sides of the San Jacinto fault near Anza and provides for exceptionally low-loss and homogeneous transmission paths (by California standards), and consequently high accuracy in determining locations and source parameters.

The ANZA network became operational on 1 October 1982, when eight of the ten stations began delivering data to IGPP/UCSD. In December 1989, the data logging system was upgraded with new equipment which enhanced the capabilities of the network. The improvements include remote control of gain and calibration circuits at each station as well as synchronous sampling of all stations in the network. The network implemented 24-bit A/D converters in 1993 and multiple sample rates in 1994. The current configuration of the ANZA network will allow for on-scale recording of local events with magnitudes less than M ~ 5. At present, over 15,000 events have been recorded during the 15 years of continuous operation.

During the last several years, significant progress has been made towards integrating the data streams of the ANZA network with the real-time processing systems of the Southern California Seismic Network (SCSN). Through the joint efforts of the personnel at the USGS in Pasadena and Caltech, we first developed a system which sent phase picks and event waveforms to the SCSN. After testing and evaluating this procedure, we determined that although it worked well, the optimal solution would be to have a direct data feed without the delays caused by intermediate processing steps. During the past year, we implemented the Object Ring Buffer (ORB) real-time software developed by the University of Colorado supported by funding from IRIS. To effect rapid data transfer to the SCSN, we have installed an ORB server on a computer at Caltech and we are now transferring all the ANZA data within ten seconds of real-time. In this way, the broadband data can be seamlessly integrated in the SCSN real-time data processing system.

The ANZA network enhances the broadband coverage provided by the SCSN in southernmost California. ANZA stations are designed to operate in remote areas without any supporting infrastructure such as AC power, telephone or computer communications. Each station can operate using solar power and all communications between stations and the IGPP are dedicated VHF, spread spectrum, or microwave radio links. Two ANZA stations (BZN and WMC) have developed significant cultural noise over the past several years. Following discussions with Dr. Jim Mori and Dr. Egill Hauksson from the SCSN, we have decided to redeploy these stations towards the south and east of the existing ANZA station;, most likely in the Borrego Valley at the current SCSN single component analog sites, COY and YAQ.

In the past year, we have installed a station on Mt. Soledad near La Jolla (see ANZA station map). This station provides extended broadband coverage to San Diego county and the offshore region complementing the nearest TERRAscope stations at Barrett Junction (BAR) and Mt. Palomar (PLM), east and north of San Diego respectively. In another project, a set of borehole accelerometers have been installed next to the Thornton Hospital on the UCSD campus. The data from these sensors will be included into the ANZA real-time processing system and transmitted to the SCSN. In addition, two new stations will be provided from internal institutional funds over the next two years.

The southern California region has generated nearly 50 magnitude 6 or greater earthquakes since 1850 (Ellsworth, 1990). Sixty percent of these moderate to large earthquakes are associated with the San Andreas and San Jacinto fault systems and their continuations into Baja California. It is interesting to note that only seven of these events have significant surface rupture. These events include the 1857 Fort Tejon (Mw = 7.8) along the Cholame and Mojave segments of the San Andreas, the 1940 (Mw = 6.9) and 1979 (Mw = 6.4) on the Imperial Fault, the 1968 (Mw = 6.5) Borrego Mountain and 1987 (Mw = 6.5) Superstition Hills located on the southern San Jacinto fault, and the 1952 (Mw = 7.5) Kern County and the 1992 (Mw = 7.8) Landers which are not directly associated with the San Andreas-San Jacinto fault system. These historical surface ruptures are shown in this figure which also highlights the two major sections without significant surface offsets: the San Bernardino and Coachella Valley segments of the San Andreas fault and the San Bernardino, San Jacinto Valley, Anza, and the Coyote Creek segments of the San Jacinto fault.

The San Jacinto fault zone is one of the most active strike-slip faults in southern California. The long-term slip rate is 1 cm/year, determined from 29 kilometers of offset of geologic formations across the fault in the last 3 million years (Sharp, 1967). Recent measurements of offset sediments in the Anza Valley yield a similar slip rate (Rockwell, et al. 1990). The Anza segment of the San Jacinto fault zone has been identified by Thatcher et al. (1975) as a seismic slip gap for a 6 M 7 earthquake. The study of Sanders and Kanamori (1984) revealed a 15 km element of the estimated seismic gap that has been virtually aseismic in modern times. Klinger and Rockwell (1989) trenched the San Jacinto Fault at Hog Lake located in the center of the Anza seismic gap and found evidence for surface rupture from three events since 1210. Additional evidence suggests that these events occurred about 1210, 1530, and 1750.

In 1988, the Working Group on California Earthquake Probabilities (USGS Open File Report 88-398) defined the Anza segment to be the 50 km section between the southern end of the inferred 1899 M=6.4 (Abe, 1988), 1918 M=6.8 (Ellsworth, 1990) rupture just north of Anza and the north end of the 1968 Borrego Mountain M=6.8 surface rupture (see Anza seismic slip gap map). They used a slip rate of 11 mm/yr, a recurrence interval of 142 years, and assumed the previous event in this segment was 1892. Based on this information a probability of 0.3 was assigned for a magnitude 7 earthquake in the Anza area in the next 30 years.

Recently, the Southern California Earthquake Center presented its Phase II report which reassesses the results of the 1988 report. Using the results of Klinger and Rockwell (1989) and Rockwell et al. (1990), the Anza segment of the San Jacinto fault zone is considered by the Working Group on California Earthquake Probabilities [1995] to be the entire 90 km long Clark fault with an average repeat time for a magnitude 7.0 to 7.5 to be 250 (+321, -145) years. Because the dimension of the segment increased, the characteristic slip is now 3.0 m (slip map).

The most significant recent information to be developed for the seismic potential of the Anza segment is the 1750 date for the last major earthquake. Using the 142 year recurrence interval of the 1988 report a magnitude 7 earthquake is now 100 years overdue. If one prefers the Phase II report then the characteristic earthquake can be a magnitude 7.5 with the peak in the conditional probability distribution in the year 2000. In either scenario, the characteristic earthquake can generate significant damage in the major population areas of the Los Angeles basin (90-150 km distant), San Diego (90 km), and the San Bernardino Valley (60 km) (map of southern California). In similar situations, significant damage was caused in San Francisco at 120 km distance by the magnitude 6.9 1989 Loma Prieta and in various parts of the Los Angeles basin by the magnitude 6.7 1994 Northridge earthquake over 100 km from the source.

Since the installation of the ANZA network in 1982, there have been eight earthquakes in southern California with magnitudes 6.0 or greater. The ANZA network recorded seven of these mainshocks, the exception being the 1986 Mw = 6.2 North Palm Springs event. Numerous aftershocks from each of these events were recorded on scale and in the cases of the 1987 Elmore Ranch and Superstition Hills events and the 1992 Landers and Joshua Tree earthquakes, foreshocks were recorded as well (map of significant events).
The evolution of the ANZA network instrumentation has greatly increased the quality of the data from regional and teleseismic events. During the 1992 Joshua Tree-Landers-Big Bear sequence, when the ANZA stations used short-period sensors with 16-bit dynamic range, only the events with magnitudes less than 5.5 were unclipped. After the 1993 upgrade to broadband sensors with 24-bit dynamic range, the 17 January 1994, Mw = 6.8, Northridge earthquake was recorded on-scale . Another interesting example of the broadband capability for recording teleseismic events by the ANZA network is shown in for the 9 June 1994, mb = 7.8, Bolivian earthquake which occurred at a distance of 66° and a depth of 600 km. For comparison, an example of a local Ml = 3.5 event is shown.

Smaller earthquakes along the San Jacinto fault zone have a strong tendency to occur in one of four clusters of activity . These clusters have in general been persistent seismic features of the entire fifteen-year operational period, but with systematic variations within clusters. The Cahuilla cluster, which is ~ 15 km west of the trace of the San Jacinto fault, has shallow seismicity, less than 6 km from the surface. The Hot Springs cluster at the north end of the array lies between the mapped traces of the Hot Springs faults at depths of 15 to 22 km. The Table Mountain/Toro Peak cluster is a more diffuse zone of seismicity that spans the trifurcation of the San Jacinto fault into the Buck Ridge and Coyote Canyon faults, and the seismicity ranges from about 7 to 17 km deep. There are a few events along the trace of the San Jacinto, e.g., a smaller cluster right beneath the town of Anza; however, the dominant pattern of activity lies off the main trace of the fault. Each of these clusters has produced at least one magnitude 4 event during the operational period of the ANZA network.

The current ANZA digital telemetry system (see general map and see data flowchart) is designed to accommodate up to 32 three-component, remote, digital seismic stations with a sampling rates of 100, 40, and 1 samples per second per component. At present, the data from nine stations are transmitted via VHF digital radio link to a relay station on the 2655-m summit of Toro Peak. A tenth station (TRO) is located on the summit of Toro and is connected to the system by wireline telemetry. An eleventh station is located on Mt. Soledad in La Jolla which telemeters data to Toro Peak using spread spectrum radios. On Toro Peak, the data are combined into a single serial channel and transmitted over a microwave link to Mt. Soledad in La Jolla and thence to IGPP at Scripps Institution of Oceanography. All components of the system have a battery backup power system or an uninteruptible power supply to minimize the possibility of losing data.

One of the major upgrades being proposed by many seismic networks is the installation of broadband sensors for source studies, structure studies, and recording of large regional events. To achieve this goal the USGS, as part of their support for the Southern California Earthquake Center (SCEC) , purchased Streckeisen STS-2 seismometers during 1991 to replace the HS-10 seismometers in the ANZA network. Most of these seismometers were deployed in December 1992 and at present, all stations except TRO have STS-2 seismometers. These broadband sensors have a corner frequency of 0.0083 Hz as compared to the 2.0 Hz corner frequency of the original HS-10 seismometers.

In the original ANZA system design a 16 bit A/D converter was used which sampled each channel at 250 samples/second. The aggregate bit rate being transmitted by each remote station was 19.5 Kbps (kilobits per second). This serial bit stream was fed to the VHF transmitter where it frequency-modulated the 216-Mhz carrier.
The original digitizers were replaced in December 1989 with a new system based on a microprocessor. These digitizers have two significant design modifications. The first was the implementation of a central time and command radio signal. This command link provided the capability to synchronously sample all stations which greatly reduced telemetry errors caused by the previous asynchronous oscillators. Also, it became possible to modify individual station parameters, such as the gain settings and calibration mode, by radio command. The other design change was to the telemetry data format, which is now composed of buffers containing one second of data for all seismic inputs, reducing the aggregate data rate by 40 percent. Each buffer has a header with diagnostic information about the telemetry system to monitor the radio transmission quality, the state of health of the digitizer, and the battery voltage.

In June 1993, the network stations were upgraded from 16-bit to 24-bit A/D converters. The 24-bit components for this upgrade were funded by the Cecil H. and Ida M. Green Foundation for the Earth Sciences. As a result of this change the ANZA stations now have identical hardware to the REFTEK RT72A-08. Any future hardware improvements developed by REFTEK or the IRIS PASSCAL program will be easily incorporated into the ANZA network stations.

Simultaneous developments in hardware and firmware resulted in improvements to the ANZA network in 1993. The hardware modification consisted of converting the dataloggers to use the standard REFTEK RT72A-08 platform which includes a 24-bit A/D converter. This modification was made possible through funds provided by the Cecil H. and Ida M. Green Foundation for the Earth Sciences. We developed and tested new algorithms for data compression and for data packet error detection and retransmission which are used in the IRIS/JSP sponsored Kyrgyz Regional Seismic Network, the Piñon Flat broadband array, and the Geyokcha small aperture array in Turkmenistan. These algorithms were incorporated into the ANZA system when the hardware was upgraded. The net result of these changes is an increase in the potential capability of the ANZA network and an extremely low data error and loss rate. The current telemetry bandwidth limitations of the existing system will allow at least 32 total stations to be recorded.

In May 1994, through IGPP and private funds, we received Digital Signal Processing (DSP) cards for each station. These DSP cards allowed us to implement a continuously recorded 40 sps data stream for the broadband sensors in addition to the present 100 sps triggered data stream.

The terminus for the VHF portion of the telemetry is on Toro Peak. A wireline telemetry bit stream from a local seismic station (TRO) is received along with the VHF radio signals from the remote stations. The network time is provided by a single GPS synchronized clock located on Toro Peak. The time data is broadcast to all the remote stations on a VHF radio link. Each of the one-second data buffers for each station-channel is received by the DATA CONCENTRATOR which sends all the data packets on a simplex microwave link to IGPP using a standard SDLC packet format. The microwave data link utilizes Motorola Starpoint equipment operating in the 1.7 GHz government band. A 1-watt transmitter at Toro Peak, feeding a 2.7-m dish antenna, provides a 34 dB signal-to-noise ratio over the 112 km path to Mt. Soledad in La Jolla where an identical antenna receives the signal. The link from Mt. Soledad to IGPP, 3.2 km away, uses a 0.1 watt transmitter and a pair of 0.7-m dish antennae. The current maximum bit rate for the microwave system is approximately 400 Kbps.

At IGPP, the microwave signal from Mt. Soledad is fed to the ENET Engine which converts the SDLC packets from the microwave link into TCP/IP packets which are received by a SUN workstation . The real-time processing system is based on the Object Ring Buffer (ORB) developed for the IRIS Broadband Array by colleagues at the University of Colorado as part of the PASSCAL program. This system is a complete real-time system which includes estimating P and S wave arrival times, event detection in multiple frequency bands, event triggers, location and magnitude estimation, and data archiving. The existing ORB system has ring buffer capacity of over 24 hours of data from the ten-station network.
In a parallel ORB process, all waveform data are copied via the INTERNET to the SCSN data center where the ANZA data are included in the SCSN real time event association, location, and magnitude estimation processing. The ANZA waveform data is received at Caltech by the SCSN between 5 and 10 seconds of real time. Real-time data feeds are being set up for San Diego State University and are currently operational to the University of Colorado.

The routine processing occurs on a weekly basis. The first step in the data processing is to remove false triggers from the data set. The seismic triggers are immediately copied to two DLT tapes for off-line storage, one which resides at UCSD and the second is sent to the IRIS Data Management Center in Seattle, Washington. The next step is to review the automatic P and S phase picks from all events and to retrieve the arrival picks from the SCSN via INTERNET. Both the hypocenter and the focal mechanism for all events are calculated from the merged ANZA-SCSN data set. Standard spectral source parameters are calculated for all events within 50 km of the network. Teleseismic phases are associated with the REB (which is also updated daily via the INTERNET) and PDE catalogs. Finally, these parameters are all stored in a permanent on-line Datascope relational database.

Fault Zone Trapped Waves

Recently, Li, et al. (1997) showed fault-zone guided waves recorded at the seismic arrays deployed above the Hot Springs cluster in the San Jacinto fault zone (SJFZ) near Anza. Three linear arrays were deployed, two on the Casa Loma strand and one on the Hot Springs strand, observing microearthquakes occurring within the fault zone. The guided wavetrains characterized by relatively large amplitudes and long period following S-waves were observed only when both the stations and events were located within or close to the fault. The amplitude spectra of guided waves showed peaks at frequencies of 4 to 6 Hz, which decreased sharply with distance from the fault. We interpreted that the fault-zone guided modes were formed due to coherent multiple reflections at the boundaries between the low-velocity fault zone and the high-velocity wall rock.

We further found that the significant fault-zone guided waves were only registered at the seismic arrays across the Casa Loma fault (CLF) which is the southern strand of the SJFZ northwest of Anza, but not at the array deployed across the Hot Springs fault (HSF) which is the northern strand of the SJFZ. This suggests that a low-velocity waveguide exists on the southern fault strand, but not at the northern fault strand. The locations of events for which we observed fault-zone guided waves suggest that this waveguide extends about 30 km along the CLF/SJF between the towns of Hemet and Anza. Since the deepest event for which we observed fault-zone guided waves at the CLF occurred at the depth of about 18 km, we interpret that the waveguide extends to 18 km depth, which is consistent with the floor of the seismogenic layer in this region. The data also show that the waveguide on the CLF dips northeastward at 75-800 while it becomes nearly vertical in the Anza slip gap. The precise earthquake locations provided by the ANZA network were essential for the interpretations of this study.

We have finished a tomographic inversion for 3-dimensional P wave velocity structure (Scott, 1992; Scott, et al. 1993). We simultaneously relocated events using the progressive location technique of Pavlis and Booker, (1980) and using 3-dimensional ray-tracing (Um and Thurber, 1987). The dataset used in the inversion consists of 564 local events recorded by the ANZA network and the Caltech-USGS seismic network. In all, 9320 very impulsive P-wave arrival times were used (quality 0 or 1 picks only, with picking error .019 and .031 seconds, respectively). Because the inverse problem is nonunique, we chose to find the smoothest model that fit the data to the accuracy of the picking error. Station corrections were included in the free parameters to account for near-surface velocity variations smaller than the grid spacing.

The P wave model shows a steep gradient at the edge of the fault zone with overall faster velocities on the northeast side of the fault. At 3-6 km depth, the signature of the fault zone is evident in the lower velocities beneath the surface trace of the fault. However, at 9 km depth, higher seismic velocities are found extending into the fault zone from the northeast block. It is interesting that this higher velocity region occurs where there is a distinct lack of seismicity on the fault. There is also a localized feature in the southwest of the modeled region, with velocities 3% slower than average, that is more than 10 km from the main trace of the fault. This area is also characterized by the presence of many hot springs at the surface which suggests the presence of cracked and broken rock at elevated temperatures.

Our success with the P-wave inversion encouraged us to attempt an S-wave inversion though we found it more difficult to obtain very high quality S picks even with the 3-component recording of the ANZA network (Scott, 1992). The S-wave dataset consisted of 3900 arrivals from 382 events. The resulting model has poorer resolution, but resembles the 3-D P-wave velocity model. In particular, a region of high velocity extending into the fault zone at 9 km depth again seems to be associated with an aseismic zone.

In a recent study conducted with Los Alamos and New Mexico Institute of Mining and Technology (Hartse, et al., 1993), we found that the regional stress field obtained through inverting focal mechanisms of small earthquakes occurring near the Anza seismic gap generally agrees with other stress models previously found for the region. The maximum compressive stress orientation is essentially horizontal and trends N 4±5 °E . The minimum compressive stress orientation is also nearly horizontal and trends N 94 ±5 °E . The intermediate compressive stress orientationis vertical to within approximately 5 ° (all 95% confidence intervals). Results from focal mechanism inversions for small geographic bins suggest that stress orientations and relative stress magnitudes show detectable heterogeneity.

In the Cahuilla swarm region , maximum and minimum compressive stress orientations appear to be rotated clockwise by about 25° relative to the regional model. In the Toro Peak swarm region horizontal stresses appear to be rotated counter-clockwise by about 10° relative to the regional model. The most abrupt changes in faulting and stress are seen in the immediate vicinity of the seismicity gap. Within the seismicity gap, the maximum compressive stress orientation appears to be oriented at 74±13° to the fault strike, in contrast with a 50° to 60° orientation along the more seismically active segments of the fault zone. Increased fault-normal compression within the gap may induce a local increase in the effective strength of the fault and thus provides a possible explanation for depressed seismic and creep activity. Alternatively, a strong, locked fault segment at the seismicity gap may have caused the observed local rotation in the stress field. The increasing depth of the seismogenic zone northwest of the seismic gap correlates well with locally increased reverse faulting and a decrease in heat flow, while southeast of the seismic gap the decrease in depth of the brittle-ductile transition correlates with an increase in heat flow and a local increase in normal faulting.

The similar waveforms for some Anza event pairs permit very accurate relative timing of phase arrivals. These times can be used for high-precision event relocations (e.g., Fremont and Malone, 1987; Xie et al., 1991; Antolik et al., 1991), and as constraints on any temporal variability of seismic velocities between the sources and receivers. Studies of this type have generally shown velocity stability; however Poupinet et al. (1984) and Ellsworth et al. (1987) documented intriguing time shifts in the coda of northern California earthquake pairs and Verwoerd et al. (1992) found time shifts in arrivals for event pairs which spanned the time of the Loma Prieta earthquake. Analysis of similar earthquakes provides a powerful tool to evaluate claims of temporal changes in velocity or coda- Q before major earthquakes, because the biasing effects of non-stationarity in the earthquake catalog can be eliminated (e.g., differences in earthquake locations and mechanisms). Similar earthquake studies have found no evidence for precursory changes in velocity (Verwoerd et al. 1992) or coda- Q (Got et al. 1990) before major earthquakes. A recent paper by Scott et al. (1994) used ANZA data to study the temporal variations of velocity in the region of the Anza gap. They found no evidence for measurable changes in seismic velocity in the Anza region since 1982.

Data from the Anza seismic network have several advantages in studying similar earthquake pairs. The network has operated for over 10 years, resulting in a catalog of over 12,000 events for analysis. Data are recorded at IGPP with a common time base for all stations. Data prior to 1989 have relative timing accurate only to within half a sample; however digitizing since 1989 has been synchronized, permitting more precise timing. All stations are three-component, so the horizontals can be used for S-wave timing and for evaluating the possibility of temporal changes in shear-wave splitting (Aster et al. 1990). Events at Anza do not tend to occur in clusters of nearly identical events (Aster, 1992), as appears to be the case at Parkfield (Antolik et al. 1991). However, we have identified a set of very similar events over the last 12 years which are suitable for cross-correlation analysis.

In two recent studies, ANZA data has been used for detailed examination of foreshocks and mainshocks. Mori (1994) used the high dynamic range of the ANZA network to examine the rupture directivity and slip distribution of the Ml 4.3 foreshock to the 1992 Joshua Tree earthquake. In this study he used a Ml 2.4 event as an empirical Green's function to deconvolve the far-field displacement pulses which were inverted for the slip distribution. He was able to show that the foreshock hypocenter was several hundred meters to the south of Joshua Tree mainshock hypocenter, and the rupture propagated to the north towards the mainshock hypocenter. Two recent papers (Ellsworth and Beroza, 1995; Beroza and Ellsworth, 1996) have utilized nine events recorded on the ANZA network with magnitudes as low as 3.5 to study the seismic evidence for an earthquake nucleation phase.

Despite the fact that the majority of Anza stations are "hard rock" sites on a granitic batholith, the recorded waveforms seem to be highly dependent on the ray path and receiver site effects, so much so that waveforms at two stations of the array are usually incoherent. During 1985, Vernon et al. (1991) conducted an experiment with a 9-element 500 m aperture array at the site of one of the ANZA array stations. In this initial study we found that details of the seismic body wave spectrum above 15 Hz for P-waves and above 10 Hz for S-waves are controlled by local site effects and spectra can be significantly different for receiver spacings as small as 50 meters. This was a surprising result for Piñon Flat Observatory, an area that has relatively little variation in geology or topography, where depth to the base of the weathered layer varies by less than 20 meters. In the spring of 1990, a 52-element high frequency small aperture array was operated at Piñon Flat Observatory for a duration of six weeks. This experiment had an aperture of 256 m and a minimum sensor spacing of 7 m (compared to 50 m for the previous experiment). The coherent part of the spectrum as a function of interstation distance for P-wave signals is below 20 Hz. for distances greater than 50 meters and below 15 Hz for the SH signals. The results of this study complement the original results of Vernon et al. (1991).

Fortunately, there is reason to believe that wave propagation at depth will be coherent over longer length scales given that lithostatic pressure closes pore spaces and cracks in the rock. Thus, events separated by several hundred meters may still show significant correlation at a single station as found by Hutchings and Wu (1990) for San Fernando aftershocks. Some evidence for longer coherence lengths in competent rock was also found by Vernon (1989), who observed a significant increase in the coherence between sensors located at 150 and 300 m depth in a borehole when compared to surface sensors spaced 150 m apart.

We have conducted a preliminary examination of the coherence of earthquake P waveforms from nearly collocated hypocenters recorded at each station. We examined pairs of earthquakes from the Cahuilla cluster with separations of less than one kilometer. The inter-event distances were determined by a relative event location procedure. The coherence estimates for each 50 meter interval were averaged together. The remarkable aspect of these results are the range of distance and frequency where there are significant coherence levels. The station BZN exhibits significant coherence above 25 Hz at event separations up to 1 km. Another interesting feature of these data are that the results vary from station to station. However, in all cases we see more coherent results than those obtained from surface arrays.

We feel that continuity of the ANZA array operations is very important. The arguments for the ANZA project remain as strong today as they were 15 years ago. The San Jacinto fault zone, with its branches and extensions into the Imperial Valley, remains one of the most active fault zones in California, and the Anza seismic gap thus remains one of the most probable sites for a moderate to major earthquake in the next few years. We are just now accumulating the amount of high quality data necessary to answer critical questions about time variability of seismicity, source mechanisms, and wave velocities. These data are essential to establishing baseline values for analysis of any future premonitory phenomena.

We believe that this network provides the highest quality data and data return rate. The ANZA system uses real-time 24-bit broadband telemetry and can easily accommodate three additional strong motion channels, which could be utilized to exploit available bandwidth. Toro Peak, the central telemetry point, is uniquely situated to provide complete coverage for the whole Coachella segment of the San Andreas fault in addition to most of the San Jacinto fault zone. These fault sections pose two of the most prominent seismic hazards in southern California. From an urban hazards viewpoint, the existing ANZA stations and line-of-sight sites available for potential future stations provide coverage for San Diego, Riverside, and Imperial Counties with a population of 3.8 million people.

In keeping with the spirit of cooperation which has characterized seismological research at Anza, and consistent with the scientific motivations of the ANZA network since its establishment, we will coordinate our research and operations effort with the parallel work conducted by SCSN seismologists (Dr. Jim Mori, Dr. Egill Hauksson, Dr. Lucy Jones, and Dr. Rob Clayton) and with SCEC.

The focus of this proposal is directed towards seismic network operations. We intend to continue to monitor regional and local seismicity and to provide on-scale high dynamic range recordings of moderate to large earthquakes. We also plan to continue to develop and improve the state-of-the-art for 24-bit real-time telemetry from remote sites adjacent to potential major earthquakes sources.
We propose the following developments for the ANZA network:

Field Operations
The data acquisition system of the ANZA network has improved considerably since the original network installation in 1982. Our experience in developing seismic networks and arrays based on the design of the ANZA system has directly influenced two upgrades to the ANZA network. In 1998-2000 we propose to continuously operate the following field configuration:
1. Eleven 3-component broadband stations.
2. All stations using Streckeisen STS-2 broadband seismometers.
3. 24-bit A/D resolution on all broadband channels.
4. 100-sps data for triggered data streams.
5. 40-sps data for continuous data streams.
6. 1-sps data for continuous data streams
7. To move BZN and WMC stations to the south and east to enhance the combined broadband coverage of ANZA and the SCSN.

1. A new project sponsored by the University of California Cooperative Lab/Campus project has installed a set of borehole accelerometers at the Thornton Hospital on the UCSD campus. The datalogging system is based on the ANZA model complete with the ORB real-time software. These data will be integrated into the ANZA dataflow and transmitted to the SCSN on a real-time basis. The station maintenance of this site is covered by the CLC program. There is no additional cost to incorporate these data into the ANZA real-time system.
2. IGPP has decided to provide equipment for new ANZA stations at a rate of one station per year for the next two years. Specific sites have not been determined yet, but the siting plan will be coordinated with the SCSN to provide the most significant enhancement to the broadband coverage for the southern California region.
At present the ANZA stations will saturate on any event inside the array which has M > 5. This limitation also was clearly apparent for large off-array events during the Landers sequence where all events M > 6.0 saturated. These large local events would provide essential data to the scientific and engineering community if they are recorded on-scale. The current dataloggers have the capability to record additional three channels which can be used to record strong motion sensors. With data compression, all VHF and microwave telemetry links have enough bandwidth to accommodate these extra data. We are not asking for funds under this proposal to implement this option, however we will attempt to generate the necessary funds from internal Institute resources and private sector support during the duration of this proposal.

From 1 October 1982 through 21 July 1997, the catalog includes waveforms and associated parameters from more than 15,000 earthquakes. We use an on-line Datascope database to store all source and waveform parameters along with pointers to easily access the waveform data.

During the period of this proposal we plan to accomplish the following:
1. Continue real-time transmission of complete ANZA waveform and parameter data via INTERNET to the SCSN.
2. Continue to keep a complete and concurrent archive of the ANZA waveform data at the IRIS DMC.
3. Determine routine locations and source parameters for all events and store the results in the Datascope relational database.
4. Archive all new ANZA waveforms on the EPOCH optical mass storage system and provide INTERNET access to this data.
5. Develop software for easier access to earthquake source and waveform parameters in Datascope relational database.

The complete waveform data set which consists of over 15,000 events is stored on-line in our EPOCH optical disk mass storage. This data is stored in the standard CSS 3.0 format complete with instrument responses and is accessible over the INTERNET. A data request is satisfied by placing the data in our anonymous FTP directory for retrieval via the INTERNET or by sending a tape copy. At present we can provide data in the following formats: CSS 3.0, SAC, or SEED. The IRIS Data Management Center is maintaining a copy of our data archive which is updated on a weekly basis.

We are currently developing a world-wide-web home-page for the ANZA network, http://eqinfo.ucsd.edu/, which provides maps and information about our database, stations, hardware configurations. In the future it will be possible to display ANZA waveforms through this interface and we plan to provide interactive access to the waveform database in the future.

The primary users of our data and results will be the general public and San Diego based media through our www homepage, our education and outreach real-time seismic displays in IGPP, at the Stephan Birch Aquarium at the Scripps Institution of Oceanography, and at San Diego State University. Additionally, researchers from academia and industry have complete access to all ANZA data and results directly through UCSD or can access data through the SCEC , Datacenter or the IRIS DMC.

The major coordination effort is with the Southern California Seismic Network. We are delivering all the ANZA network data in real-time so that the ANZA data can be combined with all the SCSN data to produce earthquake locations and magnitudes based on both datasets. To minimize confusion, the SCSN will maintain a master catalog which they will submit to the composite earthquake catalog of the Council of the National Seismic System. We are also working with the SCSN to coordinate relocating two ANZA stations to improve the broadband coverage in San Diego, Riverside, and Imperial Counties. As new equipment becomes available, we will coordinate new station deployments to optimize the broadband coverage for southern California.

The ANZA network will be providing real-time data to San Diego State University for use in their educational program and media presentations.

During 1997-1998, Dr. Yong-Gang Li from USC and Dr. F. Vernon from UCSD are collaborating on an independent field study of the fault zone guided waves to delineate the Buck Ridge, Clark Lake, and Coyote Creek faults at depth. Hypocenters based on the ANZA data are essential for this study since the SCSN has extremely poor depth resolution in this region.

The continuing operation of the ANZA Seismic Network is important to our Institute in several ways. Firstly, it provides a mechanism to have a real-time view of the local regional and teleseismic seismicity. This is important for our interactions with the San Diego news media when large earthquakes occur and in other situations where public information is needed. San Diego is the 6th largest city in the United States. Secondly, the network is important as an educational tool. Four PhDs at IGPP have been based on ANZA data as well as at least two from other universities. At present, we have one graduate student who is using ANZA data in her thesis work. More than ten undergraduate students have participated in data processing and data analysis over the years and several of those currently work in earthquake research or engineering. Finally, the ANZA model has spawned several projects, including the IRIS/JSP Kyrgyz Regional Network and small aperture arrays which are currently operating in central Asia. In turn, the ANZA project directly benefits from these other projects since the developments made for these additional systems are reincorporated into the ANZA system.

Frank L. Vernon; Associate Research Geophysicist
Glen Offield; Development Engineer
Marina Glushko; Programmer Analyst
Jennifer Eakins; Staff Reseach Associate

There are substantial resources in equipment and personnel available at IGPP and UCSD which are available for this project. We have a network of Sun and HP workstations and mass storage devices including hard disks, a 350 Gbyte erasable-optical Epoch mass storage device and a 2.5 terabyte Metrum mass storage device, and Exabyte, DAT, and DLT tape readers. There are many seismologists at IGPP who are experienced at processing and manipulating large seismic data sets, and we have considerable software designed for this purpose including the Datascope Application Package developed at the University of Colorado.

Al-Shurki, H., G. L. Pavlis, F. L. Vernon (1995). Site Effect Observations from Broadband Arrays Bull. Seismol. Soc. Amer. 85, 1758-1769.
Aster, R.C., P.M. Shearer (1992). Initial shear wave particle motion and stress constraints at the Anza Seismic Network, Geophys. J. Int., 108, 740-748.
Aster, R. C., J. Scott (1993). Comprehensive characterization of waveform similarity in microearthquake data sets, Bull. Seismol. Soc. Am., 83, 13071314.
Aster, R.C., G. Slad, J. Henton, and M. Antolik (1996). Differential analysis of coda Q using similar microearthquakes in seismic gaps; Part 1, Techniques and application to seismograms recorded in the Anza seismic gap, Bull. Seismol. Soc. Am., 86, 868-889.
Beroza, G. C., W. Ellsworth (1996). Properties of the seismic nucleation phase, Tectonophysics, 261, 209-227.
Benz, H. M., A. Frankel, D. M. Boore (1997). Regional Lg Attenuation for the Continental United States, Bull. Seismol. Soc. Am., 87, 606-619.
Ellsworth, W.L., Beroza, G.C. (1995). Seismic evidence for an earthquake nucleation phase. Science, 268, 851-5.
Haase, J.S., P. M. Shearer and R.C. Aster (1995). Constraints on temporal variations in velocity near Anza, California, from analysis of similar event pairs, Bull. Seismol. Soc. Am., 85, 194-206.
Hanson, J.A., J.B. Minster, and F.L. Vernon (1993). Characterization of Frequency Dependent Polarizations for Local Events Recorded on Surface and Borehole Instruments. Proceedings - 15th Annual Seismic ResearchSymposium. 15, 132.
Hartse, H., R. Aster, M. Fehler, J. Scott, and F. Vernon (1994). Small-scale stress heterogeneity in the Anza seismic gap, Southern California. J. Geophys. Res. 99, 6801-6818.
Li, Y. G., K. Aki, and F. L. Vernon (1997). San Jacinto fault-zone guided waves: A discrimination for recent active strands near Anza, California. J. Geophys. Res., 102, 11689-11702.
Lin, C.H., S. W. Roecker (1996). P-wave backazimuth anomalies observed by a small-aperture seismic array at Pinyon Flat, Southern California; implications for structure and source location?, Bull. Seismol. Soc. Am., 86, 470-476.
Mori, J. (1993). Fault Plane Determinations for Three Small Earthquakes Along the San Jacinto Fault, California: Search for Cross Faults, J. Geophys. Res, 98, 17,71117,723.
Mori, J. (1996). Rupture directivity and slip distribution of the M 4.3 foreshock to 1992 Joshua Tree earthquake, Southern California Bull. Seismol. Soc. Am., 86, 805-810.
Steidl, J. H., A. G. Tumarkin and R. J. Archuleta (1996). What is a Reference Site?, Bull. Seismol. Soc. Am., 86, 1733-1748.
Scott, J.S., T.G. Masters, and F.L. Vernon (1994). Three-dimensional velocity structure of the San Jacinto fault zone near Anza, California- I. P-waves. Geophys. J. Int. 119, 611-626.

Antolik, M., W. Foxall, A. Michelini and T. McEvilly (1991). Micro-earthquake clusters at Parkfield: constraints on fault-zone structure and failure processes (abstract), EOS Trans. AGU . 72, 483.
Aster, R., P. Shearer, and J. Berger (1990). Quantitative measurements of shear-wave polarizations at the ANZA seismic network, Southern California implications for shear-wave splitting and earthquake prediction , J. Geophys. Res. 95, 1244912474.
Berger, J., L.M. Baker, J.N. Brune, J.B. Fletcher, T.C. Hanks, and F.L. Vernon, III (1984). The Anza array: a high-dynamic range, broadband digitally radio telemetered seismic array, Bull. Seismol. Soc. Am., 74, 14691481.
Berger, J., H.K. Eissler, F.L. Vernon, I.L. Nersesov, M.B. Gokhberg, O.A. Stolyrov, and N.T.Tarasov (1988). Studies of High-Frequency Seismic Noise in Eastern Kazakhstan , Bull. Seismol. Soc. Am. ,78, 17441758.
Ellsworth, W.L. (1990). Earthquake history, 1769-1989, U.S.G.S. Open File Report 1515.
Ellsworth, W.L., L.D. Dietz, J. Frechet and G. Poupinet (1987). Preliminary results of the temporal stability of coda waves in central California from high-precision measurements of characteristic earthquakes, U.S.G.S. Open File Report , 591, 440460.
Fletcher, J., L. Haar, T.C. Hanks, L. Baker, F.L. Vernon~III, J. Berger, and J.N. Brune (1987). The digital array at Anza, California: Processing and initial interpretation of source parameters, J. Geophys. Res., 92, 369382.
Fremont, M-J., and S. D. Malone (1987). High Precision Relative Locations of Earthquakes at Mount St. Helens, Washington, J. Geophys. Res., 92, 1022310236.
Got, J.L., G. Poupinet and J. Frechet (1990). Changes in source and site effects compared to coda $Q^-1$ temporal variations using microearthquakes doublets in California , Pure Appl. Geophys. 134, 195228.
Gurrola, H., J. B. Minster, H. Given, F.L. Vernon, J. Berger, and R. Aster (1984). Analysis of High-Frequency Seismic Noise in the Western United States and Eastern Kazakhstan, Bull. Seismol. Soc. Am., 80, 951970.
Hartse, H., R. Aster, M. Fehler, J. Scott, and F. Vernon (1994). Small-scale stress heterogeneity in the Anza seismic gap, southern California, J. Geophys. Res, 99, 68016818.
Hutchings, L., and F. Wu (1990). Empirical Green's Functions From Small Earthquakes: A Waveform Study of Locally Recorded Aftershocks of the 1971 San Fernando Earthquake, J. Geophys. Res., 95, 11871214.
Klinger, R. E., and T. K. Rockwell (1989). Recurrent late Holocene Faulting at Hog Lake in the Anza seismic gap, San Jacinto fault zone, Southern California GSA, Cord. Sect., Abs., 42, 102.
Owens, T. J., G. L. Pavlis, and F. L. Vernon (1990). The 1990 Pinyon Flat Tight Grid Passive Source Experiment An IRIS Eurasian Seismic Studies Program Project, IRIS Newsletter, 9, 2-6.
Pavlis, G. L., and J. R. Booker (1980). The mixed discrete-continuous inverse problem: Application to the simultaneous determination of earthquake hypocenters and velocity structure, J. Geophys. Res., 85, 48014810.
Poupinet, G., W.L. Ellworth and J. Frechet (1984). Monitoring velocity variations in the crust using earthquake doublets: an application to the Calaveras Fault, California, J. Geophys. Res., 89, 57195731.
Rockwell, T., C. Loughman, and P. Merifield (1990). Late Quaternery Rate of Slip Along the San Jacinto Fault Zone Near Anza, Southern California, J. Geophys. Res., 95, 85938606.
Sanders, C.O., and H. Kanamori (1984). A seismotectonic analysis of the Anza seismic gap, San Jacinto fault zone, southern California, J. Geophys. Res., 89, 58735890.
Scott, J.S. (1992). Microearthquake Studies in the Anza Seismic Gap, Ph. D. Thesis, U.C. San Diego.
Sharp, R.V. (1967). San Jacinto fault zone in the Peninsular ranges of southern California, Geol. Soc. Am. Bull., 78, 705730.
Thatcher, W., J.A. Hileman, and T. C. Hanks (1975). Seismic slip distribution along the San Jacinto fault zone, southern California, and its implications, Geol. Soc. Amer. Bull., 86, 11401146.
Um, J., and C.H. Thurber (1987). A fast algorithm for two-point seismic ray tracing, Bull. Seis. Soc. Am., 77, 972986.
Vernon III, F.L. (1989). Analysis of data recorded on the ANZA seismic network, Ph. D. Thesis, U.C. San Diego.
Vernon III, F.L., J. Fletcher, L. Haar, A. Chave, and E. Sembera (1991). Coherence of seismic body waves from local events as recorded by a small aperture array, J. Geophys. Res., 96, 11981-11996.
Verwoerd, M.C., W.L. Ellsworth, A.T. Cole and G.C. Beroza (1992). Changes in crustal wave propagation properties associated with the 1989 Loma Prieta, California earthquake: implications for coda-Q as an earthquake precursor and for the mechanism of stress-induced velocity changes (abstract), Seismol. Res. Lett., 63, 71.
Xie, J., Z. Liu, R.B. Herrmann and E. Cranswick (1991). Source processes of three aftershocks of the 1983 Goodnow, New York, earthquake: images of small symmetric ruptures, Bull. Seis. Soc. Am., 81, 818843.

URL: http://eqinfo.ucsd.edu/proposals/AZ_prop.html