Chapter 2: Anza Seismic Network: Description
ABSTRACT
A small aperture seismic network has been operated from October 1, 1982, through the present along a 30-km stretch of the San Jacinto fault in the vicinity of the town of Anza, California. The network was installed specifically to study the scaling laws of body-wave spectra, the character of high-frequency ground motion, the physical interpretation of seismic stress drops, and the interaction of earthquakes. This region was chosen for these studies due to its high rate of seismic activity in the 2 ≤ M ≤ 4.5 range, the likelihood of a M > 6 event in the next decade, and the existence of the Peninsular Ranges Batholith on either side of the fault, reducing problems associated with attenuation and large scale anisotropy. These studies employ an instrument package with a bandwidth from 1 to 100 Hz, a dynamic range of 138 dB, a sampling rate of 250 times per second per component, and a 16-bit A/D converter. The network consists of 10 three-component stations, telemetered via digital VHF radio to a nearby mountain peak and relayed via a microwave link to La Jolla, California. A minicomputer system monitors the network's performance, detects events, and records data upon demand.
The network recorded 2239 events between network startup and February 18, 1989. About 44% of the archived events have had source parameters calculated. This network demonstrated the feasibility of digital data transmission for seismic methods with an inherent increase in data quality over analog systems.
1. Introduction
As part of the National Earthquake Hazards Reduction Program, the Institute of Geophysics and Planetary Physics (IGPP) of the University of California at San Diego (UCSD) and the United States Geological Survey (USGS) established a digital seismic network in the vicinity of the town of Anza in Riverside County, California.
This cooperative project originated in 1978 with the submission of two independent proposals to install a seismic network in the Anza seismic slip gap [gThatcher et al.g, 1975] (see also Chapter I). One proposal was sent to the USGS external grants program by Dr. J.N. Brune and Dr. J. Berger of IGPP as a part of a project to study crustal deformation at the Pinyon Flat Observatory. The other proposal was submitted by Dr. T. Hanks of the USGS to their internal grants program. The following year, the interested parties decided to utilize the strengths of each organization by creating a cooperative USGS-IGPP research project. Funding started early 1980 and, after an intensive development phase, the network became operational October 1, 1982. The primary contributors for the installation and operation of this network from the USGS are researchers Dr. Tom Hanks and Dr. Joe Fletcher, computer software specialist Larry Baker, and data analyst Linda Haar. The participating staff from IGPP includes principal investigators Dr. Jon Berger and Dr. James Brune, along with graduate student Jennifer Scott, and field and laboratory technician James Batti.
This network provides high dynamic range, broadband seismic data on about 100 earthquakes a year in the magnitude range of 2 ≤ M ≤ 4 that occur along the approximately 40-km stretch of the San Jacinto fault within the closure of this network (Figure 1a). At larger magnitudes, coverage is provided by the network of strong-motion accelerographs (shown in Figure 1b) along the San Jacinto fault, originally installed by the California Institute of Technology (CIT) and the USGS. These stations are now maintained by the USGS.
The nature and location of this network were chosen to satisfy several basic and applied research goals, as well as several additional factors specific to the vicinity of Anza. Most importantly, the multiple, near-field, digital seismograms with a bandwidth of some 100 Hz allow us to determine the locations, mechanisms, and source parameters at a level of detail hitherto not possible for earthquakes in this range of magnitude. Scientifically, these data serve two broad purposes, the first relating to earthquake ground motion scaling studies, and the second relating to investigations of how earthquakes interact with one another.
The San Jacinto fault zone in the vicinity of the town of Anza has several features that are desirable in the context of the scientific goals outlined above. The first of these is the seismic slip gap that exists in this area for M ≥ 6 earthquakes since 1890. Described originally by gThatcher et alg. [1975] and discussed in Chapter I, there is fairly good evidence that all of the San Jacinto fault zone has broken in a series of 6 ≤ M < 7 since 1890 with the exception of two segments near Anza and San Bernardino. Both of these gaps are, however, presently active at the M ≤ 5 level. This is the second reason for choosing the Anza area to deploy our network: within or near the closure of the network, we expect about one M ∼ 4 event each year (Figure 1b). There is also a 15-km seismicity gap just north of the town of Anza, nested within the longer seismic slip gap noted above [gSanders and Kanamorig, 1984]. This gap is includely aseismic at the present thresholds of detectability of the USGS/CIT Southern California Seismic Network.
The third reason for chosing this specific area is the nearby Cecil and Ida Green Pinon Flat Observatory (PFO) at which a host of geophysical observations have been made over the past 18 years. Current instrumentation at the observatory which will supplement the seismic data collected by this network include three-component long-base strainmeters, long-base liquid level tiltmeters, an array of borehole tiltmeters and strainmeters, as well as a variety of seismic instruments. Finally, the Peninsular Ranges batholith is widely exposed on both sides of the fault zone near Anza (Figure 2), and several studies have indicated that the propagation paths from local earthquakes to the stations of our network have a very low attenuation. Thus, we believed this area would have a relatively transparent "window" through which to view the earthquakes, and this would simplify the task of separating source from path effects.
2. Field Instrumentation
The Anza system is designed to accommodate data from up to 16 three-component, remote, digital seismic stations. Each station transmits information continuously using a VHF digital radio link to a relay station on top of Toro Peak (2657 m). There, the data from all stations are combined into a single serial channel and transmitted on a microwave link to Mt. Soledad in La Jolla and relayed to IGPP at Scripps Institution of Oceanography (SIO). Figure 3 illustrates the telemetry network. The station locations which were determined by Jennifer Scott of IGPP using GPS satellite receivers are shown in Figure 1b and listed in Table 1. The inter-station distances and azimuths are given in Appendix 3.
Site selection
The Anza seismic slip gap is bounded on either side by crystalline plutonic and metamorphic basement rocks. The plutonic rocks have granitic compositions and are thought to underlie the whole region [gFraserg, 1931; Dibblee, 1971; 1981], while the metamorphic rocks are roof pendants and septa formed during the emplacement of the granitic plutons.
The subsurface geology is not well constrained. It is assumed that the geology is approximately homogeneous for seismic velocities, to the depth of the seismicity which is approximately 20 km below the surface. The surface geology and the structure of the San Jacinto fault zone in the vicinity of Anza was carefully described by gSharpg [1967]. Most of the surface geology consists of basement rocks while the rest is shallow sediments and alluvium. On the northeast side of the fault the sediments are concentrated in the Garner Valley and the adjacent Santa Rosa Indian Reservation. Maximum depths for these sediments, of 200 and 250 meters respectively, were estimated from gravity surveys and water well logs [gDurbing, 1975; gBuono et al.g, 1979]. The depth of the sediments on the other side of the San Jacinto fault varies from 0 to 168 meters in the Anza, Cahuilla, and Terwilliger Valleys and may reach 250 meters along the fault trace [gMoyleg, 1976].
The topography of the area has considerable relief. The San Jacinto and the Santa Rosa Mountains are the dominant features on the northeast side of the fault. Since the remote seismic stations require line of sight for the VHF telemetry to Toro Peak, there are only limited areas where stations can be located on the north side of the fault. The topography is fairly flat near the town of Anza which allowed greater flexibility for finding sites on the southwest side of the fault. Southeast of the trifurcation of the San Jacinto fault is an extremely rugged region of deep canyons and sharp ridges with very limited accessibility.
Figure 2 is a reproduction of a geological map [gRogersg, 1966] of the Anza area with the locations of the Anza seismic stations indicated with blue dots. This map is oriented parallel to the fault system with the cardinal directions being indicated by the boundaries of each township and range. The surface exposures of the plutonic rocks are shown by the pink sections on Figure 2 while the metamorphic rocks are identified by several shades of green. The sedimentary and alluvial rocks are marked by shades of yellow. The topography is shown by the red contours.
The choice of the site for each Anza seismic station was based on the following criteria:
- line of sight to Toro Peak
- availability of 110 Volt power
- maximizing area of coverage
- surface outcrops of crystalline basement rock
Since the installation of the network, lower prices and increased availability of solar panels have eliminated the 110 V power requirement. Appendix 1 gives a brief description of the local geology and topography at each site.
Seismic stations
The two primary design considerations for these stations were to be able to record, on scale, local earthquakes with ML ≤ 4. The bigger events will generate larger ground motion, so a primary consideration for the choice of seismometer was a large case-to-coil motion. The 2-Hz GeoSpace HS-10 seismometer was selected which provides, in addition to a 1.27-cm mass motion, a separate calibration coil, field adjustable period, and a proven design dating back to the L.A.S.A. network in the 1960's. Table 2 gives the seismometer design parameters. These seismometers come in either vertical or horizontal configurations and exhibit no spurious resonances below 100 Hz. The generator constants range from 2400 to 3200 volts/m s-1. The free period and motor constant for each sensor was measured before installation and the damping was adjusted to 0.7 of critical.
The seismic velocity information is the input to a Reftek RT 24 digitizer which provides amplification, filtering, analog-to-digital conversion, and outputs a serial data stream. Figure 4 shows a block diagram for the remote station. The three seismic signals are amplified with a field adjustable gain from 24 to 66 dB in 6 dB increments (Table 3). These signals are then filtered by a 6-pole low-pass Butterworth filter which originally had a corner frequency of 100 Hz which was later moved down to 62.5 Hz. The system response for both filter corner frequencies is shown in Figure 5. Each filter and amplifier was tested for temperature stability and calibrated for absolute gain values. All the filters had amplitude and phase responses which differed by less than 1% from 0.1 to 125 Hz over a temperature range of -10C to 50C. The poles and zeros of the transfer function are given in Appendix 2.
Each channel is digitized at a sample rate of 250 samples/sec by a 16-bit analog-to-digital (A/D) converter so there is a maximum 96 dB range at each gain setting. Actually, the A/D converters are only guaranteed to be linear for 14 bits of resolution over the temperature range of -25 to +85C. The full scale limits of the A/D in the RT 24 are ± 10 Volts so the digitization constant is 3276.7 counts/volt. The system gain for all the filters and A/D converters matched, within 1%, the theoretical design values. The system noise characteristics, with a resistive input termination, have a white spectrum with at most 2 bits of noise.
| Switch | System Gain (ms-1 per) | Setting (db) | Gain (V m-1s) | Least Count | Full Scale (ms-1) |
|---|---|---|---|---|---|
| 0 | 24.1 | 16 | 4.85 x 103 | 6.29 x 10-8 | 2.06 x 10-3 |
| 1 | 30.1 | 32 | 9.70 x 103 | 3.15 x 10-8 | 1.03 x 10-3 |
| 2 | 36.1 | 64 | 1.94 x 104 | 1.57 x 10-8 | 5.15 x 10-4 |
| 3 | 42.1 | 128 | 3.88 x 104 | 7.87 x 10-9 | 2.58 x 10-4 |
| 4 | 48.2 | 256 | 7.76 x 104 | 3.93 x 10-9 | 1.29 x 10-4 |
| 5 | 54.2 | 572 | 1.55 x 105 | 1.97 x 10-9 | 6.44 x 10-5 |
| 6 | 60.2 | 1024 | 3.10 x 105 | 9.83 x 10-10 | 3.22 x 10-5 |
| 7 | 66.2 | 2048 | 6.21 x 105 | 4.92 x 10-10 | 1.61 x 10-5 |
The 16 data bits for each sample are combined with two-component identification bits, a battery status bit, a parity bit and several frame synchronizing start and stop bits to form a 26 bit data word. Each channel is sampled in order, and the digital information is sent out in a serial line. The aggregate bit rate which is transmitted from each remote station is 19.5 Kbps (kilobits per second). The serial output of the RT 24 is then sent to a Reftek RT 21 VHF transmitter which frequency-modulates a carrier in the 216-220 MHz band. The modulator uses a frequency-shift-key encoding with a ± 5 KHz deviation from the center frequency. The frequency bandwidth which is occupied by most of the energy in the VHF transmission is about 24 KHz. Probably the most difficult components to troubleshoot were the VHF transmitters and receivers. The original transmitter design was not stable for seasonal temperature changes, and the receivers initially had a nonlinear phase response over the necessary 0.5 to 10 KHz modulation bandwidth. These problems were resolved in late 1983.
The three seismometers at each station are fastened to a 17" x 17" x 34" aluminum plate by mounts which are machined to be orthogonal to a precision of ± 0.1. The plate is then leveled and oriented on three inch studs which are embedded in a concrete pad. The pad is poured, directly on crystalline rock if possible (stations KNW, RDM, TRO), or is dug into the weathering layer and poured over rebar stakes which help couple the pad to the ground. The seismometers, the digitizer and the transmitter are housed inside a water-tight metal enclosure and are protected by concrete walls and a inch plywood top. Outside the metal enclosure, but inside the concrete walls, is a 12-volt storage battery which powers the station. The battery is float-charged by a power supply connected to a 110 V, 60 Hz, A.C. powerline at all stations. At station CRY, a 53-watt ARCO M55 solar panel has also been used successfully to power the equipment. The power consumption of the entire apparatus is about 8 watts.
Relay station
The terminus for the VHF telemetry is on Toro Peak (Figure 6). A bit stream from a local seismic station (TRO) is received along with the data from the nine remote stations. Each one of these asynchronous serial bit streams is input to a Reftek RT 44 remote interface unit where it is converted into parallel words one sample wide and momentarily stored. A station identification code is added to the data word along with a parity error bit indicative of the quality of the VHF transmission path from the remote station to Toro Peak. These data words are then sampled in turn by a commutation which selects each receiver three times in 4 msec. The multiplexed samples, which are 31 bits long, are then converted into a single bit stream which has an aggregate bit rate of 240 Kbps.
Microwave link
The microwave data link utilizes Motorola Starpoint equipment operating in the 1.7 GHz government band. A one-watt transmitter on 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 (Figure 3) where an identical antenna receives the signal. From there, the information is relayed directly to IGPP, 3.2 km away, using a 0.1 watt transmitter and a pair of 0.7 m dish antennas. On average, the down time of the network due to atmospheric conditions should be less than 1 hour per year.
As shown in Figure 6, the bit stream from the RT44 digital multiplexer at the relay station is fed directly onto the baseband of the microwave transmitter. Its baseband frequency response is ± 0.5 dB from 4 to 408 KHz. The maximum bit rate from the RT44, with the design 16 stations in operation, will be approximately 336 Kbps. Most of the power in this bit stream is confined to the pass band of the microwave, so that it is easy to reconstruct the bit pattern at the receiver end. To increase the number of stations in the network beyond 16 and keep the per station bit rate, either a wider bandwidth microwave transmitter can be installed or a more efficient data packing scheme can be used.
3. The Detection and Recording System
The detection and recording system is composed of two major elements: a special-purpose microcomputer and a general-purpose minicomputer. The design philosophy was to minimize interface hardware and software and isolate it in a custom microcomputer which as far as the minicomputer is concerned, is simply a standard peripheral. The minicomputer, with its real-time multi-task operating system, provides the power and flexibility to allow the event detection and data recording software to be written entirely in Fortran.
Telemetry interface unit
At IGPP, the microwave signal from Mt. Soledad is fed to the Telemetry Interface Unit (TIU) which is built around an Adaptive Sciences Inc. Modulus-One 6809 based microcomputer. Its job is principally to relieve the data logging computer of the task of servicing interrupts that occur for each data word received. With 10 three-component stations, these interrupts occur on average every 133 μsec, but on occasion as often as every 83 μsec. Secondarily, the TIU is assigned the task of attaching a clock time to the data. Each of the remote stations operates on an independent clock. The relay station of Toro Peak also has its own clock which controls the multiplexing of the remote stations. The time of year of the data samples is noted only at the TIU by reference to a single WWVB synchronized clock.
The TIU receives the telemetry serial bit stream, frames or synchronizes it, and performs a serial to parallel conversion. At this point, the data word consists of 32 bits for each data channel: 16 data bits and 16 channel and status bits.
The TIU demultiplexes this data stream into one of two data buffers, attaching the time of the first scan from the WWVB clock to each buffer. As each buffer fills, the TIU interrupts the minicomputer and a data transfer is initiated. This scheme of using two buffers allows efficient direct memory access into the minicomputer, transferring a block of data about 4 K words at a time. Each buffer will contain the data for a 500-msec interval. Thus, there will be 125 samples for each of the 30 channels plus a three-word time header. Each element in the data buffers is initialized to plus full scale before the buffers are filled to serve as a flag for a missing datum. Thus, for example, if the microwave link is lost momentarily all data for that time interval will be full scale. If for some reason a remote station's VHF link goes "down," the storage registers at the relay station on Toro Peak will retain the last good datum received and keep transferring that to the microwave link. Thus, the data from that channel should be of constant value for the duration of the VHF link "down" time, while the other two channels of data from the affected station will be full scale.
The data logging system.
Fundamental to the design and operation of the ANZA network was the development of the data logging software by Larry Baker of the USGS. The following description of the software was originally presented by gBerger et al.g [1984]. The data logging software is composed of six major processes running in a time-sharing manner on a Digital Equipment Corp. PDP 11/34 minicomputer using the RSX-11M operating system. They are:
- the sampler, which handles the interrupts from the TIU and transfers data blocks into memory
- the trigger, which examines a specified subset of data and determines if a data block is to be recorded
- the delay line manager, which separates overlapping events into distinct data packages, duplicating data as necessary
- the buffer manager, which manages the memory blocks of data and routes them to the various processes as necessary
- the output writer, which writes data blocks to magnetic tape, disk or both when so directed by the trigger
- the message writer, which handles the system messages to the console device and disk files.
The software was designed with three principal requirements in mind: speed of execution, conservation of memory resources, and tolerance of transient error conditions. The first requirement is governed by the basic machine speed and the aggregate design input data rate of 240 Kbps. The second requirement stems from the PDP 11/34 memory capacity of 256 Kbyte. As it takes some time to decide if an event has occurred, and as one wants a pre-event sample of the data, the entire data set must be stored for at least the sum of these two periods. For a relatively dense network like ANZA where the interstation travel time is on the order of a second, one can contemplate storing this entire data set in memory rather than on disk. The software, however, has been designed to handle either storage method.
The direct memory access method of interfacing the TIU to the PDP-11/34 causes very little processing load on the data logging machine, thus maximizing the time available for other processing. Several blocks of memory are available in the PDP-11/34 to receive the incoming data, and the blocks can be filled in any order. Once a block has been filled, it is stamped with an internal sequence number (independent of the WWVB time stamp), and routed to a trigger program and delay line program simultaneously. If an event is detected, the trigger program directs the delay line program to send the sequence of data blocks on to an output writer program, which then actually writes the data to a disk file or magnetic tape file. Otherwise, after a certain latency period, the data blocks are returned to the free-block pool where they become available to be filled again. It is the responsibility of the delay line program to guarantee that the proper sequence of data is sent to the output writer(s), regardless of how the delay line itself is managed (e.g., in memory or using a temporary disk file).
The task of detecting seismic events in the data telemetered from the Anza Array is, in theory, not difficult since we are not trying to record small earthquakes whose energy barely exceeds ground noise. Further, as our stations are closely spaced, we can use a "voting" algorithm to improve the ratio of detections to false alarms. In fact, the principal challenge is not so much to detect events as it is to reject transient noise of high energy that may occur due to telemetry errors. Thus, the algorithm is based more on a knowledge of the "noise" than the signal: the telemetry drop-out data pattern described in the section "Telemetry Interface Unit" is used to advantage.
A certain subset of data channels are specified as trigger channels (typically, all vertical components) and another subset are monitor channels. The detection algorithm processes these channels only. As with most seismic detectors, the trigger algorithm compares the ratio of a short-term to long-term average of an estimate of the signal variance to a constant-the trigger level-[gBerger and Saxg, 1981]. For those channels designated as trigger channels, when this trigger level is exceeded, a flag is set which remains in that condition for a specified interval to allow for the seismic travel time across the network. If, at any instant, more than a certain number of trigger channels are in the "set" condition, an "event" is declared. The delay line manager is then instructed to record the current data block, several previous blocks (a preevent leader), and a header. After the event is declared over, several more data blocks are recorded to provide a post event trailer.
For speed, all required components of the system (including the entire delay line) are permanently resident in the computer's memory, whether they are busy or idle. Thus, there is no delay for servicing any requests due to disk loads or other operating system overhead. The other programs, such as the status display or trigger parameter modification programs, are loaded from disk as needed, and can share the same memory space. For memory conservation, more than half of the code in each program is shared with every other program via a common in-core system library. This reduction in program storage requirements frees memory that is used for additional data buffers and contributes to the overall robustness of the system. The existing system has a memory capacity of over 5 sec of data from the design 16-station network.
The task of event detection and recording requires almost all of the capabilities of the PDP-11/34 system. The requirement of at least 8 sec of network data memory resident, and the systems hardware limitation of 256 kB, means that there is little room for other tasks. Further expansion of the network will require more memory or a disc resident data buffer. Software maintenance and development can only be performed on the system when the detection and recording tasks are not operating. Normally, the data analysis and system development is done on a similar system either in the La Jolla labs or at the U.S.G.S. facilities in Menlo Park.
Data processing
When a field data tape is completed, the first step in evaluating the events is to plot all triggers. All earthquakes are then copied onto two archive tapes, one which is sent to our colleagues at the USGS at Menlo Park and one which is kept at IGPP. The USGS picks arrival times and calculates spectrum parameters, then returns these values to IGPP, as well as the hypocenters and source parameter estimates of each event. This information is then put in a relational data base which is used to isolate appropriate events for specific research projects.
4. Coherence Between Seismometers
The ANZA seismic network was the first 16-bit digital telemetry network ever installed. A series of tests were designed to verify the quality and consistency of the data recorded on this system. The two primary questions to be answered were whether ground noise can be resolved across the whole frequency band and whether sensors will give the same output from the same input. A special mounting plate was designed and machined a pair of horizontal seismometers, or a pair of vertical seismometers, could be mounted parallel to each other. Four horizontal and two vertical seismometers were tested at high gains to observe if background noise was coherent. If the noise is coherent between parallel seismometers then it is likely that ground noise, not instrument noise, is being measured.
Figure 7 shows the coherence of the ground noise measured on two parallel horizontal seismometers. The top graph gives the power spectrum for one of the two input series. The middle and bottom graphs show the phase and magnitude squared coherence (MSC) between the two series respectively. The MSC shows a high level of coherence up to about 80 Hz. However, since it is not unity for this band, the intrinsic noise of the seismometer is approaching the level of the ground noise. The loss of coherence at the highest frequencies is an effect of the anti-alias filters which start filtering seismic data at 62.5 Hz. The 95% confidence level for the MSC being significantly different from zero is marked by the dashed line. Since each data channel is sampled by the same analog to digital converter, offset by 1/750 second, the phase response is not equal to zero for the coherent band, instead it has a constant linear trend to 90 Hz where the signal becomes incoherent. The results from other components are similar to these.
After the sensor ground noise coherence measurements were completed, the amplifier gains were reduced to the minimum possible value (24db), and the station was left with a pair of vertical seismometers for 15 days. During that time a local ML ∼ 2.5 event was recorded. Figure 8 shows the results of the power spectrum, phase, and MSC for this seismogram. The results are similar to the ground noise, except that MSC is essentially unity until 60 Hz. and starts degrading between 60 and 80 Hz. The loss of coherence starts 10 Hz lower for the large dynamic range signal of the earthquake. This implies a nonlinearity above 60 Hz in the seismometers for large amplitude events. The results of these tests indicate that our seismic recording system has a linear response for wide dynamic range signals and that they are able to resolve ground noise.
5. Ground Noise
The network was operated at high amplifier gains (54 db) for several short periods of time to record events with low magnitudes (1 < ML < 2.5). At this gain level, the ground noise at each site can be resolved. The maximum, minimum, and median power spectra using data from nine ANZA stations are plotted in Figure 9. The power spectral levels are higher than previous results in the same frequency band from Kazakhstan, USSR, and Nevada, USA [gGurrola, et alg., 1989]. The minimum power levels for the ANZA network have similar values to the maximum levels recorded with low wind velocities at the Nevada sites. Values for ground noise power spectra from the Kazakh sites are lower than those from Nevada. The differences can probably be attributed to the vault construction techniques used by each network. (Refer to gBerger et al.g [1988] and Gurrola et al. [1989]).
6. ANZA Data
The intent of the design for the ANZA network was to record, on scale, a large range of earthquake magnitudes. During the period of time from the initiation of data logging in October 1982 until February 1989 there were 2293 earthquakes which were archived. A time line of the cumulative events (Figure 10) shows a fairly constant rate of occurrence with occasional jumps for various swarms and aftershock sequences. Of the events prior to July 1988, 738 were near enough to the network, with sufficient amplitude, for the calculation of seismic source parameters. These events have shear wave moments from 1.3 x 1017 to 1.2 x 1022 dyne-cm and have a magnitude range of 0.6 ≤ ML ≤ 3.9. Under the current configuration, with a minimum gain of 24 db, the dynamic range for most of the stations in the ANZA network will saturate on events which have magnitudes ML > 4.0. If the minimum gain level were to be reduced to 0 db then it may be possible to record events with ML ≤ 5. The catalog of earthquakes recorded on the ANZA network includes many events from outside the network. Figure 11 shows all the events recorded by the network which were also located by Southern California seismic network [gNorris et al.g, 1986; gGiven et al.g, 1987]. The dashed box is the same boundary which defines network area in Figure 1b. The prominent clusters outside the network include the North Palm Springs sequence which started in 1986 and the Elmore Ranch-Superstition Hills earthquakes and aftershocks. Other notable events such as the Oceanside sequence, the Whittier Narrows mainshock, several of the San Diego Bay events, and the largest of the Coalinga events were also recorded. Unfortunately, the low frequency response of the seismometers eliminates most teleseismic events from being recorded during the normal low gain network operation. The ANZA network was fully operational during both the 1984 Borah Peak (M = 7.3) and for the 1985 Michoacan (M = 8.3) earthquakes and did not trigger. However, this gap in our recording capability can be eliminated by using sensors with a better gain response between 0.1 and 2 Hz.
7. Conclusions
The ANZA seismic network is the first operational digital telemetry network devoted to the study of earthquakes. The seismograms recorded by this network have provided high dynamic range data from small to moderate earthquakes (0.5 < ML < 4.0) across a frequency band of 1 to 100 Hz. The current database of ANZA earthquakes contains 2293 events from southern California. Source parameters have been calculated for events which are located near or inside the closure of the network. The proven design of the ANZA seismic network has been used as the basis for the NRDC nuclear test ban verification networks in the Soviet Union and the United States, as well as for the Parkfield Dense Array in Parkfield, CA. The ANZA network continues to provide a baseline of events which can be used for many studies related to earthquake prediction, earthquake source mechanics, seismic wave propagation and attenuation, and seismic velocity structure determination.
Appendix 1- Station Descriptions
| Station | Geology | Topography |
|---|---|---|
| TRO | Surface rocks: quartz diorite-granodiorite. Located in the middle of the Santa Rosa Pluton. Nearest boundary of Pluton is > 5 km away. No nearby sediments. Nearest metamorphic outcrops ∼1 km distant. [gDibbleeg, 1981] |
Top of 2657 meter peak, which is part of an elongated ridge with a northwest major axis. Elevation drops to 1700 meters ∼5 km north and ∼5km south of the station. |
| CRY | Surface rocks: small outcrop of tonalitic basement, at the end of a small alluvial valley, part of Coahuila Valley Pluton. The nearest boundary of the pluton is ∼1 km to the northwest. Shallow local alluvial sediments < 20 m thick surround the station at a minimum distance of 30 meters away. [gSharpg, 1967; Moyle, 1976] |
Located on a small knoll ∼10 meters high surrounded by granitic boulders. Small stream bed runs by to the east with a closest approach of ∼30 meters. Local terrain is quite flat. |
| KNW | Surface rocks: tonalitic basement located in the San Jacinto Mountain Intrusive Complex. The weathering layer is about 20 meters thick. The nearest sediments are ∼1km to the southwest. No metamorphic outcrops within ∼5km. [gHillg, 1981; Hill, 1984; Fletcher et al., 1989] |
Located on the crests of a small descending ridge oriented NNW. Local topography is small low amplitude ridges and valleys overlying a general slope from the San Jacinto Mountains on the north to the saddle at Mountain Center ∼2 km to the south. |
| PFO | Surface rocks: decomposed quartz diorite-granodiorite, part of the Haystack Pluton. Depth of the weathering layer is ∼20 meters. No nearby sediments. Nearest metamorphic outcrops are ∼3 km to the west and south. [gDibbleeg, 1981; Parcel, 1981; Wyatt, 1982; Fletcher et al., 1989] |
Located on Piñon Flat. The topography is flat for several km in all directions, with a ∼.5o slope to the south. Pinon Flat bounded by Asbestos Mtn.(N), Deep Canyon (E), Santa Rosa Mtn. (S), and Palm Canyon (W). |
| FRD | Surface rocks: Tonalite of the Coahuila Valley Pluton. Nearest sediments are alluvium ∼50 m to the north which reach a depth of ∼200 m about 2 km northwest of the station. Nearest metamorphic outcrops are 3 km to the south of the station. [gSharpg, 1967; Moyle, 1976] |
Located near the edge of a shallow alluvial valley and on the flank of a low east ridge trending along the southern side of Terwilliger Valley. |
| RDM | Surface rocks: small pendant of metamorphic rocks in the Coahuila Valley Pluton. The metamorphic rock primarily consist of larded gneiss. Station located several meters from the metamorphic- tonalite contact. The surface boundary of the pluton appears to be 1 km to the northeast. Nearest sediments are >1 km distant. [gSharp,g 1967] |
Located on top of a small 1394 m mountain which rises 400 m above the surrounding land. Red Mtn. is part of northwest trending ridge which extends 10 km from the 1735 m Coahuila Mtn. |
| SND | Surface rocks: decomposed tonalite of the Coahuila Valley Pluton. Nearest sediments are in a small shallow alluvial valley ∼50 m away. Nearest metamorphic outcrops are ∼100 m away. This station is located in the trifurcation area of the San Jacinto fault zone and is ∼100 m from the surface trace of the Clark fault. [gSharpg, 1967] |
Located on north end of Table Mountain. Station overlooks slope down to Hamilton Creek and Burnt Valley, which are 30 m lower than the station. |
| BZN | Surface rocks: decomposed tonalite of the Coahuila Mtn. Pluton which extends many km west of the station. Nearest sediment contact is ∼50 m east of the station. Sediments get progressively thicker towards the east until reaching a depth of 150-200 m several km away in Terwilliger valley and overlie the Coahuila Valley Pluton. The nearest metamorphic contact is 2 km to the south. [gSharpg, 1967; Moyle, 1976] |
The site is near the top of a gentle eastward dipping slope which has minimal relief. |
| WMC | Surface rocks: alluvium and sediment which are about 60 m thick. Site is ∼30 m north of a metamorphic outcrop. There are no plutonic outcrops within 2 km. Site is 3 km from the surface trace of the Clark fault. [gSharpg 1967, gMoyleg 1976, and Fletcher, personal communication, 1989] |
Topography is gently sloping to the south with no relief. Nearest topographic feature is Thomas Mtn. 4 km to the north. |
| LVA | Surface rocks: granodiorite of the Collins Valley Pluton. Nearest sediment contact is from the small, alluvium-filled Lost Valley just north of the station. The nearest metamorphic outcrop is ∼1 km to the south of the station. |
Site is on a low rise on the south side of Lost Valley, which is essentially flat. The valley is surrounded by ridges several hundred meters in height around Lost Valley. Hot Springs Peak is 3.2 km south and 500 m above the LVA station. 6 km north of LVA is the 600 meter deep Coyote Canyon. |
Appendix 2 - Instrument Response of the ANZA Network
There is one type of sensor at each site. Almost all the time the sensors are GeoSpace HS-10 seismometers which have a 2 Hz natural frequency. Occasionally, particularly in 1982-83, one station might have a set of Kinemetrics S-1 seismometers which have a natural frequency of 0.2 Hz.
All channels are sampled 250 times per second. A 6-pole Butterworth low-pass anti-aliasing filter with a corner frequency of 100 or 62.5 Hz. (which change with time) is provided by the Refraction Technology RT24 digitizing unit. The conversion factor for the A/D converter is 3276.7 counts per volt, or 70.3 dB relative to 1 count per volt. The RT24 units also allow for additional gain from 24 dB to 66 dB in steps of 6 dB.
The ANZA instrument response as a function of frequency is given by
T(s) = TS(s) TA(s) GS GD GADC
where TS is the normalized transfer function of the seismometer, TA is the normalized transfer function of the low-pass anti-aliasing filter. GS GD, and G ADC are the effective seismometer generator constant, the programmable gain in the RT24 unit, and the analog-to-digital converter respectively. GD is dimensionless and can change with time, and its value is documented in the station parameter files recorded on the data tapes. The normalized response of the seismometer is given by
TS(s) = s2 / ( s2 + 2λωos + ωo2 )
where s = 2πif, ωo = 2πfo
where fo is the natural frequency of the seismometer, and λ is the damping coefficient.For the normal damping coefficient of λ = 0.707 the two zeros of the seismometer are:
s = (-0.0 ± 0.0i)ωo
and the two poles are:
s = (-0.707107 ± 0.707107i)ωo.
The transfer function of the 6-pole Butterworth anti-aliasing filter is given by
TA(s) = ωA6 / D(s)
D(s) = (s2 + 1.931852ωAs + ωA2) (s2 + 1.414214ωAs + ωA2 ) (s2 + 0.517638ωAs + ωA2 )
where
ωA = 2πfAand fA has two possible values at separate times.
The poles of this filter are at:
s = (-0.965926 ± 0.258819i)ωA,
(-0.707107 ± .707107i)ωA,
(-0.258819 ± .965926i)ωA.
TA is dimensionless.
Appendix 3
| TRO | CRY | BZN | KNW | WMC | SND | PFO | LVA | RDM | FRD | |
|---|---|---|---|---|---|---|---|---|---|---|
| TRO | 0.00 | 29.04 | 22.55 | 33.33 | 23.79 | 17.36 | 8.95 | 24.36 | 40.54 | 16.93 |
| CRY | 29.04 | 0.00 | 10.60 | 16.33 | 5.56 | 11.68 | 26.50 | 28.68 | 12.56 | 14.61 |
| BZN | 22.55 | 10.60 | 0.00 | 24.85 | 9.27 | 8.22 | 23.33 | 18.18 | 22.96 | 5.68 |
| KNW | 33.33 | 16.33 | 24.85 | 0.00 | 15.60 | 20.16 | 26.56 | 42.50 | 15.17 | 26.10 |
| WMC | 23.79 | 5.56 | 9.27 | 15.60 | 0.00 | 6.61 | 20.94 | 27.03 | 16.92 | 11.14 |
| SND | 17.36 | 11.68 | 8.22 | 20.16 | 6.61 | 0.00 | 15.84 | 22.88 | 23.54 | 6.25 |
| PFO | 8.95 | 26.50 | 23.33 | 26.56 | 20.94 | 15.84 | 0.00 | 30.80 | 36.37 | 18.61 |
| LVA | 24.36 | 28.68 | 18.18 | 42.50 | 27.03 | 22.88 | 30.80 | 0.00 | 40.62 | 16.64 |
| RDM | 40.54 | 12.56 | 22.96 | 15.17 | 16.92 | 23.54 | 36.37 | 40.62 | 0.00 | 27.16 |
| FRD | 16.93 | 14.61 | 5.68 | 26.10 | 11.14 | 6.25 | 18.61 | 16.64 | 27.16 | 0.00 |
| TRO | CRY | BZN | KNW | WMC | SND | PFO | LVA | RDM | FRD | |
|---|---|---|---|---|---|---|---|---|---|---|
| TRO | 0 | -83 | -102 | -54 | -79 | -83 | -18 | -147 | -75 | -104 |
| CRY | 97 | 0 | 140 | 7 | 80 | 97 | 79 | 147 | -55 | 122 |
| BZN | 78 | -40 | 0 | -11 | -8 | 36 | 56 | 151 | -48 | 85 |
| KNW | 126 | -173 | 169 | 0 | 167 | 151 | 115 | 161 | -126 | 156 |
| WMC | 101 | -100 | 172 | -13 | 0 | 112 | 79 | 158 | -68 | 141 |
| SND | 97 | -83 | -144 | -29 | -68 | 0 | 66 | 170 | -68 | 172 |
| PFO | 162 | -10 | -124 | -65 | -101 | -114 | 0 | -160 | -86 | -133 |
| LVA | 33 | -33 | -29 | -19 | -22 | -10 | 20 | 0 | -40 | -11 |
| RDM | 105 | 125 | 132 | 54 | 112 | 112 | 94 | 140 | 0 | 123 |
| FRD | 76 | -58 | -95 | -24 | -39 | -8 | 47 | 169 | -57 | 0 |
References
Berger, J. and R.L. Sax, Seismic detectors: the state of the art, S-cubed, gTechnical Report, SSS-R-8088g, 1981.
Berger, J., H. Eissler, F. L. Vernon, I. L. Nersesov, M. B. Gokhberg, O. A. Stolyrov, and N. D. Tarasov, Studies of high-frequency seismic noise in Eastern Kazakhstan, gBull. Seism. Soc. Am., 78g, 1744-1758, 1988.
Buono, A, W. R. Moyle Jr., and P. Dana, Water resources of the Santa Rosa Indian Reservation and vicinity, Riverside County, California: U.S. Geol. Surv., gOpen File Rep., No. 79-1172g, 35 pp., U.S. Geol. Surv., Open-File Serv. Sect., Denver, CO.
Dibblee, T.W., Jr., Regional geologic map of San Andreas and related faults in eastern San Gabriel Mountains, San Bernardino Mountains and San Jacinto Mountains, California, U.S.G.S. Open-filed map, scale 1:125,000, 1971.
Dibblee, T. W., Geology of the San Jacinto Mountains and adjacent areas, Geology of the San Jacinto Mountains, gSouth Coast Geological Societyg, 1-47, 1981.
Durbin, T. J., Ground-water hydrology of Garner Valley, San Jacinto Mountains, California, a mathematical analysis of recharge and discharge, U.S.G.S. Open-File Report No. 75-0305, 56pp, 1975.
Fletcher, J. B., T. Fumal, H. Liu, and R. Porcella, Near-surface velocities and attenuation at two bore-holes near Anza, CA from logging data, preprint, 1989.
Fraser, D. M., Geology of San Jacinto quadrangle south of San Gorgonio Pass, California, gCalif. J. Mines and Geology, 27g, 494-540, 1931.
Given, D. D., L. K. Hutton, and L. M. Jones, gThe Southern California Network Bulletin July-December 1986g, U.S.G.S. Open File Report 87-488, 1987.
Gurrola, H., J. B. Minster, F. L. Vernon, J. Berger, H. K. Given, and R. Aster. A comparative study of high-frequency seismic noise at selected sites in the USSR and USA. gProc. I1th Annual AFGL/DARPA Seis. Res. Symp.g, 1989.
Hill, R. I., Geology of the Garner Valley and vicinity, gSouth Coast Geological Society, Annual Field Trip Guidebook, 9g, 90-99, 1981.
Hill, R. I., Petrology and petrogenesis of batholithic rocks. San Jacinto Mountains, southern California, gPh.D. dissertationg, California Institute of Technology, Pasadena, 1984.
Moyle, W. R., Geohydrology of the Anza-Terwilliger area, gU. S. Geological Survey, Water-Resources Investigations, 76-10g, 1976.
Norris, R., L. M. Jones, and L. K. Hutton, gThe Southern California Network Bulletin July 01 through December 31, 1985g, U.S.G.S. Open File Report 86-337, 1986.
Parcel, R. F., Structure and Petrology of the Santa Rosa Shear Zone in the Pinon Flat Area, Geology of the San Jacinto Mountains, gSouth Coast Geological Societyg, 139-150, 1981.
Rogers, T. H., Geologic Map of California Santa Ana Sheet, scale 1:250,000, California Division of Mines and Geology, 1966.
Sanders, C. O., and H. Kanamori, A seismotectonic analysis of the Anza seismic gap, San Jacinto fault zone, southern California, gJ. Geophys. Res., 89g(B7), 5873-5890, 1984.
Sharp, R. V., San Jacinto fault zone in the Peninsular Ranges of southern California, gGeol. Soc. Am. Bull., 78,g 705-730, 1967.
Thatcher, W., J. A. Hileman, and T. C. Hanks, Seismic slip distribution along the San Jacinto fault zone, southern California, and its implications, gGeol. Soc. Am. Bull., 86g, 1140-1146, 1975.
Wyatt, F., Displacements of surface movements, gJ. Geophys. Res., 87g, 979-989, 1982.
