Bulletin of the Seismological Society of America, Vol. 81, No. 4, pp , August 1991

Transcription

1 Bulletin of the Seismological Society of America, Vol. 81, No. 4, pp , August 1991 THE EARSS DIGITAL SEISMOGRAPH: SYSTEM DESCRIPTION AND FIELD TRIALS BY K. R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK ABSTRACT An earthquake detection and recording system known as EARSS has been developed for permanent seismograph stations and temporary field installations. It records three components of ground motion with a dynamic range of 120 db. A frequency-domain algorithm detects earthquakes and initiates the recording of data on magnetic tape. Alternatively, EARSS can record data continuously, for preselected periods of time, or recording can be triggered by a time-domain phase picker. Up to 1500 earthquakes (25.5 Mbytes) can be recorded on each magnetic tape cartridge. The field version of EARSS supplies power to the tape drive only when data is being written to tape, thus reducing the normal power consumption of 12 watts (at 12 volts) to 2.5 watts. A field trial using a network of eight EARSS seismographs resulted in 1020 successful station-days of operation from a possible total of 1098 station-days (3 years). Of the 78 lost days of operation, 23 were due to power supply problems external to EARSS, and 52 were caused by a low-temperature failure of the recording system, which has since been corrected. A total of 442 Mbytes of data were recorded, of which about 250 Mbytes were useful data. INTRODUCTION An earthquake detection and recording system has been developed at Geophysics Division, New Zealand Department of Scientific and Industrial Research (now DSIR Geology and Geophysics) for permanent seismograph stations and temporary field installations. It is known as EARSS (an acronym formed from Equipment for the Automatic Recording of Seismic Signals), and it uses many of the ideas developed in the earlier SNARE seismograph network recording system (Gledhill and Randall, 1986). EARSS was developed because no commercially available digital seismograph simultaneously fulfills all the following requirements. 1. Reliable detection of earthquakes on a single station in the background of seismic noise. This is necessary because the large amount of data generated by digital data acquisition methods means that it is not possible to record a continuous record of ground motion; nor is a large number of false triggers acceptable. 2. Recording of a wide dynamic range of ground motion amplitude. 3. Accurate measurement of three components of ground motion. 4. A large data recording capacity. 5. Low power consumption so that remote field operation is possible. EARSS is a three-channel digital seismograph that employs gain-ranging techniques to acquire data with a dynamic range of 120 db. The frequencydomain detection algorithm of Gledhill (1985) is used to identify earthquakes, which are then recorded on magnetic tape; up to 1500 earthquakes can be stored 1380

2 THE EARSS DIGITAL SEISMOGRAPH 1381 on each magnetic tape cartridge. The low-power version of EARSS supplies power to the tape drive only when data are being written to tape, thus reducing the continuous power consumption to 2.5 watts. The instrument has been in use in prototype form since early 1987, and in the first 6 months of 1989 an extended field trail was conducted using both the permanent station version of EARSS and a low-power field version. This article describes the EARSS system and reports the results of the field trials. SYSTEM DESCRIPTION The photograph of a permanent station EARSS in Figure 1 clearly shows the front panel layout. A small display area of lights and switches is used to indicate system status and control alarm conditions. Additional removable panels cover the electronics, tape drive, and radio. All connectors are on the back panel of the instrument. The portable version of EARSS uses the same front panel, but it is housed in a fiberglass case with lid for environmental protection, instead of the free-standing cabinet shown in Figure 1, and the connectors are on the sides of the case. Figure 2 shows a functional block diagram of the EARSS system. Physically, EARSS is built around three interconnection systems, referred to as buses. A data acquisition bus is used to input the signals from the sensors and condition these for conversion to digital counts. The data acquisition bus is connected to the rest of EARSS via the data I/O ports shown. The main system bus is used!~ :ii il i FIa. 1. A photograph of an EARSS showing the front panel layout.

3 1382 K. R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK BASIC EARSS SYSTEM DATA I/O ] PORTS CLOCK J P ALLE" PO TS CON~R[TROL CONTROL l PORT ACQUISITION SECTION 1 RADIO PROCESSOR TER.~MINAL TAPE DRIVE FIG. 2. A functional block diagram of the EARSS system. by the main microprocessor (the controller) to communicate with the rest of the system, and a power bus is used to control the power supply to the radio, tape drive, and other parts of the system. Two microprocessors oversee the operation of EARSS; one is the main controller, while the other (the storage processor) handles the storage of selected data on magnetic tape. Timing is provided by a real-time clock on the system bus, which uses a specially cut crystal to minimize temperature effects. Timing accuracy is maintained by comparison with hourly time signals broadcast by some New Zealand medium wave radio stations or an externally injected time signal. The internal radio is checked and controlled by one of the parallel ports. Data buffering is achieved by the use of two 16 kbyte ring-buffers. A parallel port transfers data from the controller to the selected ring-buffer, and a second parallel port is used for buffer selection. The storage processor has access to the data in the ring-buffer that is not currently selected by the controller and can communicate with the controller via another parallel port. On a signal from the controller, the storage processor reads the data from one of the ring-buffers and stores it on magnetic tape using the tape drive shown. The tape drive used is a 25 megabyte Cipher series 525 Floppy Tape cartridge tape drive (Cipher Data Products, 1984), which was extensively modified so it could be powered-down when not in use. An operator can interact with EARSS using the serial control port shown. A terminal connected to this port can display and change the operating parameters of the system. An optional parallel printer port is also available so that a permanent record of system operation can be made. The data acquisition bus allows the main microprocessor to have full control over data acquisition. A block diagram of the EARSS data acquisition section is shown in Figure 3. Three independent input channels are provided, and each

4 THE EARSS DIGITAL SEISMOGRAPH 1383 ) (BATTERY VOLTAGE) > AAF 0) I- _z >" t AAF II AAF. F DATA ACQUISITION CONTROL BUS ADC o. ~<? Fro. 3. A functional block diagram of the EARSS data acquisition section. has its own adjustable gain amplifier (AGA), anti-alias filter (AAF), and gainranging amplifier (GRA). A multiplexer is used to select each channel, and a sample and hold amplifier (SHA) is used to "freeze" the incoming signal so that it can be converted to digital counts by the analog-to-digital converter (ADC). The data are then transferred to the controller on the data acquisition bus via the data I/O ports. In normal operation, the ADC is also used to monitor the system battery voltage. Wide dynamic range operation is possible because the GRA is used to adjust the gain before each sample is acquired. The ADC is a 13-bit device (12-bit plus sign) that provides a dynamic range of 72 db. This is then extended to 120 db by the GRA. A major problem encountered in most data acquisition systems is the data recording hardware. In low-power systems that require mass storage, this problem is particularly acute. A very flexible data storage setup is implemented in EARSS to overcome these problems. The two 16-kbyte ring-buffers appear only as a data port and control port on the system bus. New samples are stored in the current ring-buffer until it is full, then the controller switches to the second ring-buffer, power is supplied to the tape drive, and the buffered data is written to tape. Meanwhile, acquired samples are stored in the second ringbuffer. By switching between the two ring-buffers, data are recorded with the "power-hungry" tape drive only taking power for a short time to record each buffer. A further benefit of this hardware ring-buffering system is that up to one buffer full of preevent data can be retained once an event trigger has been confirmed; the actual length of preevent data recorded is controlled by one of the system operating parameters. THE OPERATmN OF EARSS To explain the operation of EARSS, we will use the case when it is working in the frequency-domain trigger mode, which is the default. Periodically, the controller inputs data from the "outside world" using the data acquisition

5 1384 K.R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK section. Sampling rates ranging from 25 to 200 Hz are available on all channels. The incoming data stream is scanned by two data processing algorithms; a frequency-domain earthquake detector and a time-domain phase picker. The frequency-domain detector (Gledhill, 1985) uses the frequency and amplitude information in the incoming signal to identify earthquakes in a background of seismic and cultural noise. Once an earthquake has been detected, the correct number of samples are sent to the current ring-buffer so that the required length of preevent data can be recorded. The controller then switches to the second ring-buffer and sends a command to the storage processor, which writes the buffered data to magnetic tape. By switching between the two ring-buffers, recording continues until the detection algorithm indicates that the event is over. The phase-picking algorithm is operated in parallel with the detection algorithm, and is based on the work of Allen (1978). Individual channels can be set so that either P phases or S phases are identified. In the S-phase mode, a P pick on the vertical component is used to initiate the search for an S phase on the corresponding horizontal components. In normal operation, phase picks are only reported if the detection algorithm reports a coincident event. Using the control port, it is possible to change most of the operating parameters within EARSS. However, to simplify the use of EARSS, three primary parameters can be used to initiate the desired setup. These are the number of operational channels, the operational mode, and the detection type. The operational mode is used to specify how recording will be initiated and terminated. EARSS can be set to record continuously or to have recording controlled by the frequency-domain detector (the default), the time-domain phase picker, or the system clock. The clock can be programmed to initiate and terminate recording so that preselected periods of data can be recorded. The detection type is used to identify the earthquake type, which the frequency-domain detection algorithm will search for. Standard settings specify teleseisms, regional earthquakes, or microearthquakes. The parameters of the detection algorithm are adjusted, and the sampling rate is set to appropriate values depending on the detection type selected: 100 Hz for microearthquakes, 50 Hz for regional earthquakes, and 25 Hz for teleseisms. The system clock is battery backed, as is a small section of the system memory. The selected operating parameters are stored in the battery-backed memory and are reloaded when EARSS is next turned on. This feature allows an EARSS system to be preprogrammed for later installation. EARSS PERFORMANCE The performance characteristics of EARSS are summarized in Table 1. Tests on the gain-ranging accuracy indicated a worst case error of less than 0.5 per cent over the full range of input voltages. The advantages of gain-ranging in a short-period seismograph are demonstrated in Figure 4. The top trace is plotted at a scale to show the preevent seismic noise, while the lower trace has been scaled to show the maximum S-phase amplitude. Note that the difference in scale factor is Although gain ranging causes a reduction in absolute resolution for large signals, the resolution relative to the signal size remains constant and is set by the ADC used. Gain ranging is thus a very effective method of extending the dynamic range of a short-period (narrow-band) digital seismograph. However, gain ranging breaks down for broadband recording

6 THE EARSS DIGITAL SEISMOGRAPH 1385 Maximum number of channels Input voltage range Sensitivity (noise limit) Channel-to-channel crosstalk Maximum voltage gain Gain-ranging accuracy Analog-to-digital conversion Maximum usable dynamic range Sampling frequencies Anti-alias filtering Overall electronic response Timing accuracy Recording media Recording capacity Power suppiy Operating temperature range Operational modes Dimensions Weight TABLE 1 SPECIFICATIONS OF THE EARSS PORTABLE microvolts to 5 volts Less than 1 microvolt RMS Better than db 84 db 0.5% 12-bit plus sign 120 db 25 Hz, 50 Hz, 100 Hz, 200 Hz 0.3 x Sampling frequency (second-order Butterworth) 0.16 Hz to 0.3 x Sampling frequency (-3 db) Better than 0.02 sec absolute (including time correction) Digital magnetic tape (DC600A cartridge) hours, three channels 100 Hz (25.5 Mbytes; up to 1500 events) 12 volts at 2.5 watts 0 to 45 C (see text) Frequency-domain trigger, time-domain trigger, continuous recording, recording for pre-set periods 550 mmx 360 mm x 310 mm 10 kilograms EVENT: JUNE S Z , '$ E 48.9km X 1000 FrG. 4. A example of the advantage of gain-ranging. The top trace is plotted at a scale to show the preevent seismic noise, while the lower trace has been scaled to show the maximum S-phase amplitude. Note the scale factor difference of 1000 and the obvious danger of picking first motions on low gain traces. because a small short-period signal can be present on top of a larger long-period signal. If EARSS were to be used for recording broadband data, an ADC with higher absolute resolution would be required. Although some of the components used in the construction of EARSS are not specified for operation below 0 C, all EARSS systems are tested over the temperature range -10 to 45 C. The recommended operating temperature range for the magnetic tape cartridge used by EARSS is 5 to 45 C, but tests have given good data recovery rates down to -10 C. The EARSS electronic response was measured by applying a constant amplitude sine wave with varying frequency to the input and recording the signal at

7 1386 K. R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK two gain settings. The amplitude response of the electronics at each frequency was then estimated by taking Fourier transforms of the recorded signal. The normalized system response (for both gain settings) is plotted in Figure 5a, together with the theoretical response calculated by convolving the characteristics of the anti-alias filter and the direct current decoupling capacitor in the amplifier. As we did not have access to a shaking table to test the overall system response (EARSS plus seismometer), alternative tests were carried out on a complete seismograph including a Mark Products L4C-3D 1 Hz seismometer. The method described by Chapman et al. (1988) was used whereby a step current is applied to the calibration coil of the seismometer, and the resulting transient signals are recorded. By taking the Fourier transform of the recorded calibration pulses and using the given parameters of the input current and calibration coil, an empirical measure of the overall system response can be found. However, the step function lacks high frequencies, and there are problems caused by the mutual inductance between the calibration and pickup coils above 10 Hz, so it is necessary to derive a theoretical form for the seismometer response to extend the calibration to the higher frequencies. The seismometer response can be expressed as a function of the overall gain, the critical frequency, and the seismometer damping (e.g., Aki and Richards, 1980); estimation of these parameters at lower frequencies can be used to extend the calibration to higher frequencies. Following Chapman et al. (1988), an iterative nonlinear least-squares technique was used and the results are shown in Figures 5b and c. FIELD TRIALS The first field installation of an EARSS prototype was on the Island of Rarotonga in the Cook Islands in May The purpose of this installation was to record the T phases generated by the nuclear tests being carried out on the French Polynesian Island of Moruroa. The prototype EARSS operated for a year and successfully recorded all eight nuclear explosions detonated at Moruroa in Further prototype installations followed in New Zealand, and several problems in the microprocessor and storage electronics and software were diagnosed and corrected. Two EARSS prototypes were employed as part of a temporary network to record the aftershocks of the 1988 Te Anau (M L = 6.1, M s = 6.7) earthquake (Reyners et al., 1991). By late 1988 a production model of EARSS was ready for testing, and a joint project between Geophysics Division DSIR, and the Institute of Geophysics at Victoria University of Wellington was organized for early The objective of the field trials was to test both permanent station and portable versions of EARSS and collect earthquake data that could be used for research projects at Victoria University. Eight EARSS systems were deployed in central New Zealand from late January to early July 1989 (see Fig. 6). The operational performance of the network is shown in Figure 7. Two of the instruments (MPE and PDD) were removed ahead of schedule and used as part of a temporary network installed to record aftershocks of a large earthquake which occurred in southern New Zealand on 31 May The recording period involved a possible total of 1098 station-days of operation (see Fig. 7). However, 78 station-days of operation were lost leaving an actual total of 1020 (93 per cent). The causes of the failures are summarized in

8 THE EARSS DIGITAL SEISMOGRAPH ~ - ~ -,..., ~,, ~9 10 o < > I0-I i0-2 lo-z r i illrll i i iiiz~rl ~ i i,1=1 10 o 10' 10 z Frequency [Hz] (a) 1011 o J 9 E < 101o s / 0~ n 105 ~,,,r,,,, 10-I 10 o i i i iiijrr ~ p, rill z z 10 o z Frequency [Hz] Frequency [Hz] (b) FIG. 5. The response curves of an EARSS system. The normalized electronic response of EARSS is shown in (a); the solid line is the theoretical response curve, while the stars give the measured values. The lower diagrams show the overall response of EARSS when coupled with Mark Products L4C-3D seismometer; (b) is the amplitude response, and (c) the phase response. The solid lines show the total response of the system estimated from the corrected Fourier transform of the step function used for the calibration (dashed lines). (c) Table 2. It should be noted that during most of the trial period stations were only visited once every two weeks, so an instrument failure soon after a routine visit resulted in up to 14 days of lost operation. The cause of the power supply problems external to EARSS (23 days lost) include accidental turning-off of power and cut power cables. The largest cause of lost operating time was a recording failure due to low temperatures which has since been tracked down to an excessively temperature sensitive capacitor and corrected.

9 1388 K.R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK 40'30'S KI 41" S ~.~. CAW MTW PDD==.WEL.WDW NDE[" ;4 bs~e /. J 41 30"S 42 o S CC~ / ' 50 km EARSS Temporary Station Wellington Network Station r ~ E 1750 E 1760 E FIG. 6. A location map of the field trial network. The field trial stations are shown as solid squares; the solid circles are the stations of the telemetered Wellington Seismograph Network. During the recording period a total of 442 Mbytes of data was recorded, of which about 250 Mbytes were considered useful data and kept. The data set includes 1273 earthquakes within 100 km of the network and 299 more distant regional and teleseismic events. Regional and teleseismic events were only kept if they were recorded on at least four stations. The data collected during the operation of the network are being used to study shear-wave splitting (Gledhill, 1990), derive earthquake source parameters, investigate converted phases related to the dipping interface between the Australian and Pacific plates, which lies below the Wellington Region, and study high-frequency surface waves. DISCUSSION EARSS was primarily designed as a short-period digital seismograph to replace permanent photographic recording seismographs and portable smokedpaper instruments. At the time of writing, 20 permanent EARSS stations have been installed as part of the upgrade of the New Zealand Seismograph Network, and 10 low-power EARSS portables are in regular use. However, the flexibility inherent in the design of EARSS, particularly the bus structure and the emphasis on microprocessor control, has meant that it is easy to adapt EARSS for other geophysical data acquisition applications. EARSS systems have been used to record explosions for crustal structure studies,

10 THE EARSS DIGITAL SEISMOGRAPH 1389 BSE HHE I -"JJJJJ//'JJJJJJ//'J.-/ S S I KSE MPE l NDE PDD PRE l S TEL /J/JJJJJ /fjjjff//jjjjjjjjjj, i =, t J JANUARY FEB MARCH APRIL MAY JUNE JULY 1989 FIG. 7. The operational performance of the field trial network. The shaded bars indicate the operational periods for each station. An "S" indicates a seismometer change or resiting, and an 'T' indicates an instrument change. TABLE 2 EARSS FIELD TRIAL RESULTS Possible station-days of recording Lost station-days Power supply problems Clock board failure Low temperature problems Total Actual station-days of recording: 1098 (3 years) 23 (2%) 3 (0.3%) 52 (4.7%) 78 (7%) 1020 (93%) adapted to record marine seismic lines at sampling frequencies of up to 4000 Hz, and configured to acquire electrical resistivity data. Data collected by EARSS are being analyzed by two different systems. At the Seismological Observatory (DSIR Geology and Geophysics) a modified version of the California Institute of Technology-USGS system CUSP is being used for routine analysis. Tape images are read into a host VAX computer via a personal computer (PC). The tape images are then sorted into CUSP event files for analysis. A PC-based system has been developed at the Institute of Geophysics, Victoria University of Wellington, to analyze the data collected by microearthquake surveys conducted using EARSS portable seismographs. Recently, UNIX computers have been added, but the PC has been retained to sort EARSS tape image files into event files ready for analysis.

11 1390 K. R. GLEDHILL, M. J. RANDALL, AND M. P. CHADWICK Improvements and additions to EARSS are also underway or planned. These include a 16-channel version for telemetered seismograph networks (similar to SNARE), broadband data acquisition electronics, and the use of alternative storage media, including removeable hard disks and solid state memory. ACKNOWLEDGMENTS Allan Cresswell is largely responsible for transforming EARSS from a prototype to a production instrument. Wayne Richardson is thanked for his help during the field trials. This article has been reviewed by John Taber, Russell Robinson, and Martin Reyners, all of whom made useful comments that improved the final result. REFERENCES Aki, K. and P. G. Richards (1980). Quantitative Seismology: Theory and Methods, W. H. Freeman, San Francisco. Allen, R. V. (1978). Automatic earthquake recognition and timing from single traces, Bull. Seism. Soc. Am. 68, Chapman, M. C., J. A. Snoke, and G. A. Bollinger (1988). A procedure for calibrating short period telemetered seismograph systems, Bull. Seism. Soc. Am. 78, Cipher Data Products (1984). Series 525 Floppy Tape cartridge tape drive product description, Cipher Data Products, San Diego, California. Gledhill, K. R. (1985). An earthquake detector employing frequency domain techniques, Bull. Seism. Soc. Am. 75, Gledhill, K. R. (1990). A shear-wave polarization study in the Wellington Region, New Zealand, Geophys. Res. Lett. 17, Gledhill, K. R. and M. J. Randall (1986). SNARE: an earthquake detection and recording system for small seismograph networks, Bull. Seism. Soc. Am. 68, Reyners, M., K. R. Gledhill, and D. Waters (1991). Tearing of the subducted Australian plate during the Te Anua, New Zealand, earthquake of 1988 June 3, Geophys. J. Int. 104, DSIR GEOLOGY AND GEOPHYSICS WELLINGTON, NEW ZEALAND (K.R.G., M.J.R.) INSTITUTE OF GEOPHYSICS, RSES VICTORIA UNIVERSITY OF WELLINGTON NEW ZEALAND (M.P.C.) Manuscript received 17 September 1990