1U.S. Geological Survey.
2 Alaska Volcano Observatory, Alaska Division of Geological and Geophysical Surveys, 794 University Ave., Suite 2001, Fairbanks, AK 99709.
A system of networked personal computers provided real-time data acquisition and analysis during the 1991 eruption of Mount Pinatubo. The computers collected data telemetered from seismometers, tiltmeters, and lahar detectors. The seismic network provided earthquake location and magnitude information from the digitized event data, Real-time Seismic Amplitude Measurements and Seismic Spectral Amplitude Measurements. The seismic amplitude, seismic spectral, tiltmeter and lahar-detector data were processed automatically. Processing the seismic-event data required transferring the data from the acquisition computer to an analysis computer and then timing the events interactively. Data were sent via modem to other parties for further analysis. The modem connection also enabled scientists in the United States to troubleshoot the system. As a result of the Pinatubo experience, numerous improvements have been made to the system.
Note to readers: Figures open in separate windows. To return to the text, close the figure's window or bring the text window to the front.
The availability of low-cost, increasingly powerful microcomputers provides volcanologists with cost-effective tools for acquiring and analyzing geophysical data in real time. Lee (1989) described a system that used a personal computer to digitize signals from up to 16 seismic stations and store the records of the events. The digitized waveforms for the events were transferred via a Local-Area Network (LAN) to a second computer where they were timed and located. March and Power (1990) used such a system networked to a UNIX workstation to acquire and process seismic data during the 1989-90 Redoubt, Alaska, eruptions. Murray (1992a) described a low-data-rate telemetry system whose data can be received by a laptop computer with a 300-baud modem. This system enabled real-time acquisition and display of data from tiltmeters, temperature sensors and similar instruments. Using the data management and plotting program BOB (Murray, 1990a,b, 1992b), these data could be compared in real time with each other and seismicity as recorded by a Real-time Seismic Amplitude Measurement system (RSAM) (Endo and Murray, 1991). A mudflow-detection system developed at the Cascades Volcano Observatory (R.G. Lahusen, oral commun., 1990) during the 1989-90 Redoubt, Alaska, eruption also used a personal computer to acquire and plot the data in real time.
Together, the above items can constitute the core of a volcano observatory's data-acquisition and analysis system. However, as stand-alone systems, they do not foster interaction between different investigators or correlation of different data sets. The importance of being able to compare different data sets in near real time cannot be overstated. In reference to predictive methods used at Mount St. Helens, Swanson and others (1983) stated "the accuracy of our predictions depends on interactive use of all data by cooperating geophysicists, geologists, and geochemists." Therefore, it was necessary to integrate the stand-alone systems listed above into a system that would allow any of the computers being used for analysis to access any of the data. This was accomplished by providing paths to move data automatically from the different acquisition computers to the networked computers.
These paths consisted of both hardware (parallel-to-serial converters, the LAN, serial-port buffers) and software (data-conversion programs, modifications to existing programs). With the data stored on a network computer, each data-analysis computer, in addition to performing its specialty, can analyze the data via the network. Though all analysis could have been done on a single computer, we found that it was better to use three computers because more people could work simultaneously. One computer was dedicated to timing, locating, and archiving earthquakes. Between June 5 and June 12, 1991, this computer was in constant use at Mount Pinatubo because of the large number of earthquakes recorded. The other two computers shared the tasks of viewing the other data, report writing, outside communication, and system maintenance. Together, the computers formed the real-time volcano-monitoring data-acquisition and analysis system used by the Pinatubo Volcano Observatory (PVO).
This report describes the system that was used during the 1991 Mount Pinatubo eruptions. As a result of our experience at Mount Pinatubo, new programs have been written to further speed the processing and analysis of data. These improvements will be summarized. We wish to emphasize that any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Figure 1 shows the hardware configuration for the data acquisition and analysis system used at PVO. Four computers acquired the data. WILLIE, an IBM-AT compatible acquired the seismic-event and Seismic Spectral Amplitude Measurement (SSAM) data. RICKY and SLEEPY, both IBM-XT compatible computers, acquired data from the lahar-detector and tiltmeter networks, respectively. RSAM data were collected by a Tandy Model #100 computer (Endo and Murray, 1991).
Figure 1. The Pinatubo Volcano Observatory's computer-based data-acquisition and analysis system. This system is described in detail in the text.
At periodic intervals, the acquisition computers transmitted their data (except seismic-event data) via a print sharer to the serial port of PORKY, an IBM-AT compatible computer, where it was available for analysis. The SSAM data were transmitted every minute, tiltmeter and RSAM transmitted data every 10 min, and lahar data were transmitted whenever they were received from the field. The print sharer replaced a special circuit that allowed multiple devices to share a single serial port (Murray, 1992b). Though designed to allow multiple computers to share a single printer, in this configuration the print sharer buffered data sent by each of the acquisition computers and allowed each computer's data to be sent in turn to PORKY. Using the print sharer eliminated the need to assign each device its own transmission window during which it could transmit data without interfering with the other devices.
Three computers were used for data analysis. They accessed data on WILLIE and PORKY through a 2 megabit-per-second peer-to-peer network (Lee and others, 1989). All computers on the network were configured as servers, enabling any computer to access data or programs on any other computer. All data and programs moved between machines through the network, almost eliminating the need for the floppy disk drives, which might have quickly failed as ash from the eruption infiltrated every piece of equipment. Though each computer had a specialized task, each also served as a backup in the event that another failed. SQUAB, an 80386-based IBM-compatible computer, was used primarily for earthquake processing, analysis, and archiving. DAVE and ANDY, both IBM-AT compatible computers, were used for analysis and plotting of RSAM, SSAM, tiltmeter and lahar-detector data. They also were used for writing observatory communications and volcano advisories. A 9600-baud modem attached to ANDY provided access to the data-analysis computers from distant locations. The software package Remote (DCA Crosstalk Communications, Rosewell, Ga.) simplified connecting to the system. With Remote, the keyboard and monitor at the remote site behaved as though they were directly connected to ANDY. Commands entered on the remote keyboard cause ANDY to respond as if they were typed on its keyboard. Staff at the Cascades and Alaska Volcano Observatories were able to acquire data as well as troubleshoot the system from their offices in the United States.
At the time of the system's installation, all clocks were set to local time, not Greenwich Mean Time (G.m.t.). It was quickly discovered that the seismic acquisition program MDETECT (Tottingham and others, 1989) was time-tagging the SSAM data to the PC's clock while the event data were time-tagged to the PC's clock plus 8 hours (the time correction needed for G.m.t. in California, where the program had been written). It was several weeks before this discrepancy was resolved.
Unfortunately, at the same time that the time discrepancy was resolved, it was decided that all seismic data would be recorded in G.m.t. This was a mistake. Having to make the mental adjustments between G.m.t. and local time when correlating seismic events with other observations and talking with local officials only adds to the confusion inherent to volcanic crises. The usual argument for using G.m.t. is that the data can be easily compared with data sets collected by other groups worldwide. These comparisons are done so seldom during a crisis that this argument cannot be justified. After the crisis, a simple program can adjust the time to G.m.t. if necessary. Also, the Philippines does not seasonally adjust its local time as the United States does with daylight savings. This factor eliminates the argument to use G.m.t. in order to have a consistent time base.
Six short-period vertical seismic stations and one three-component seismic station radio-telemetered their data to PVO via a standard analog seismic telemetry system. The received data were recorded in four different ways: (1) analog drum recorders, (2) digitized waveforms of discrete events, (3) RSAM, and (4) SSAM. Each method has its limitations and strengths. The character of the seismicity, which may change through time, determined which technique was most useful. By using all four techniques, we were able to monitor seismic activity throughout the eruption. Monitoring began in early May, when there were only a few events per day, and continued through June 15, when only one station continued to transmit, and its record was largely saturated.
Signals from at least two seismometers were always being recorded on analog drum recorders. Despite advances in computer-based data acquisition, drum recorders were still the single most important technique for recording the seismic data. A glance at the drums gave a quick overall view of the current level of seismicity and its character. By displaying the records on a wall, one could qualitatively assess the change in seismicity through time. The records were also used to interpret data from other techniques, which in one way or another measure only a portion of the signal. Digitized events do not measure extended periods of tremor or constant activity. RSAM and SSAM can be contaminated with telemetry or cultural noise. The analog records, even when saturated, provided a qualitative context in which to interpret the other, more quantitative methods.
Seismic-event data were acquired and processed on the computer WILLIE using the method described by Lee and others (1989). The program MDETECT (Tottingham and others, 1989) digitized seismic signals at 100 samples per second per channel using a 12-bit analog-to-digital converter. When MDETECT detected an event, the digitized data for the event were written to disk. Periodically the event data were transferred to SQUAB, where the events where timed with the program PCEQ (Valdes, 1989) and locations and magnitudes calculated by using HYPO71PC (Lee and Valdes, 1985; 1989). After being timed and located, the data were archived to tape.
Two programs were used to plot the data. The program AcroSpin (Parker, 1990) plotted the hypocenters in a three-dimensional format that could be rotated about any axis. An early version of the program SEISPLOT (March and Murray, 1992) provided a one-page summary of earthquake activity. The summary combined an earthquake epicenter plot, a time-depth plot, and a table summarizing the number, magnitude, and depth of the earthquakes recorded (fig. 2).
Figure 2. Plot produced by the program SEISPLOT (March and Murray, 1992). In the upper left-hand corner is an epicentral map with the seismic stations noted. A time-depth plot is shown along the bottom, and in the upper right-hand corner is a table summarizing the earthquake activity with respect to depth and magnitude.
Processing the digitized events was the most time-consuming element in the entire system. Because of the large number of events recorded, SQUAB was in constant use from about June 5 to June 12, either transferring and archiving events or timing and locating them. Also, lack of integration of the timing program and the location program made it difficult to check hypocenter solutions for accuracy or to re-time them. Despite these problems, the system worked remarkably well, seldom falling behind by more than 12 hours.
The RSAM data were transferred to PORKY via the print sharer (fig. 1). A batch file was written so that a single command would cause the program BOB to quickly plot the latest RSAM data. During the period June 12 through early July (when there were insufficient seismic stations to locate events), RSAM was the most useful quantitative measure of seismicity.
The version of MDETECT used at Pinatubo had been modified to calculate 1-minute averages of the spectral content of the seismic signals in 16 selected frequency bands (Rogers and Stephens, 1991). One-minute SSAM data were both stored on WILLIE and sent to PORKY via the print sharer (fig. 1). WILLIE's printer port, instead of its serial port, was used to transmit the data because the transmission time was faster and less likely to cause timing problems with MDETECT.
SSAM data were viewed with two programs (Marso and Murray, 1991). The first, VU_FFT, was a simple GWBASIC program that produced a bar-graph representation of the data for a specified time and station (fig. 3). VU_FFT was useful to see the spectral content of a specific event or to track shifts in spectral content during periods of extended tremor or during mudflows.
The program Surfer (Golden Software, Boulder, Colo.) was used to produce daily contour plots for each station (fig. 4). The plots were placed on a wall to produce a long-term view of SSAM. Each contour plot required about 15 minutes to produce. The process was delicate; a single mistake would doom the plot with little or no indication why. However, the dramatic changes in the spectral content of the seismic signals could be clearly seen on the analog drum recorders (Power and others, this volume). This, in addition to the occasional difficulties in producing the plots, was the reason that SSAM was not a major factor in forecasting the Mount Pinatubo eruption.
Figure 3. Example plot of SSAM data for station PIEZ at selected times produced by the program VU_FFT. The first 10 columns display 1-Hz-bandwidth data centered about the frequencies (cf) shown. The columns labeled RSAM, LOW, MED, and HIGH display data from wider frequency bands: 0.7-10.0 Hz, 1.5-3.0 Hz, 3.0-4.5 Hz, and 4.5-6.0 Hz, respectively. The data are arbitrary units of the signal amplitude in each frequency band.
Figure 4. SSAM contour plot of station PIE for June 12, 1991, produced by the program Surfer. The x-axis represents time, and the y-axis represents the different frequency bands. The contours represent the amplitude of the seismic signal through time in the various bands, in the same manner that contours represent elevation on a standard topographic map. The numerical values of the contours are arbitrary units of the signal amplitude.
Data transmitted from two tiltmeters were logged on SLEEPY (Ewert and others, this volume). At 10-min intervals, the data were transmitted to PORKY via the print sharer (fig 1). Once PORKY received the data, the program BOB was used to view and compare the data with RSAM (fig. 5).
Though the telemetry system was designed to transmit data from such instruments as strainmeters, gas detectors, and ground-temperature sensors in addition to tiltmeters, lack of time prevented their installation. If installed, their data would have been processed and available for analysis in the same manner as the tiltmeters'.
Figure 5. Plot produced by the program BOB showing RSAM and tilt data from station UBO. This plot shows the slowing in radial tilt that coincided with the seismic swarm on June 7.
The lahar-detection system was first installed at PVO as a stand-alone system with no connection to other computers (Hadley and Lahusen, 1991). RICKY received, stored, and plotted the data. Later, RICKY was programmed to send the received data out its printer port and through a parallel-to-serial converter to PORKY via the print sharer (fig. 1). With the data on PORKY, analysis was done with BOB and plotted in conjunction with the RSAM and SSAM data.
From the experience gained during the Mount Pinatubo crisis, several improvements have been made to the system. Among them are the following:
The system described here performed well during the 1991 Mount Pinatubo crisis, as it provided information that was essential in issuing the eruption forecasts. The only hardware component that failed was a hard-disk drive that began to exhibit erratic behavior in early July. Before it failed completely, it was replaced by a new disk that had been purchased in Manila.
The lack of hardware failures is even more remarkable, considering that the entire system was moved four times; (1) it accompanied the initial USGS response team from the United States to the Maryland Street apartment on Clark Air Base, (2) it was moved from the apartment to the Dau Complex on Clark Air Base as unrest intensified, (3) from Dau it was moved to Clark BaseOps just after the climactic eruption, and (4) in November 1991, the system was moved off Clark Air Base to Quezon City. All moves on Clark Air Base were accomplished in a matter of hours.
Many of the limitations of the system deployed in 1991 have been addressed; the current system has been streamlined. Also, the availability of ever more powerful computers at ever decreasing prices will continue to add to the speed and capability of the system such that more sophisticated analytical programs are feasible. The system provides a model for future systems, both at established, permanent observatories or, as in the case of Mount Pinatubo, in response to an unmonitored reawakening volcano.
The system described in this report is the result of integrating the work of several different teams. The authors especially thank Willie Lee, U.S. Geological Survey, for his perseverance in the development and documentation of a PC-based seismic-event detection system; John Rogers, U.S. Geological Survey, for his assistance in modifying the program MDETECT to transmit SSAM data out the printer port; and John Lahr, U.S Geological Survey, for his suggestion of using a print-sharing device to buffer the low-frequency data transmitted to PORKY. We also acknowledge the assistance and patience of the Escuela Politecnica Nacional, Quito, Ecuador, and INGEOMINAS, Manizales and Pasto, Colombia. These groups patiently worked with early versions of the system and provided us with feedback regarding what worked well and what did not.
We also thank the Pinatubo Volcano Observatory staff; it was they who actually made the system work.
Ellis, J.O., 1989, Expanding the input multiplexer for the Data Translation, Inc. Model DT2821 analog-to-digital converter: U.S. Geological Survey Open-File Report 89-201, 5 p.
Endo, E.T., and Murray, T.L., 1991, Real-time seismic amplitude measurement (RSAM): A volcano monitoring and prediction tool: Bulletin of Volcanology, v. 53, p. 533-545.
Ewert, J.W., Lockhart, A.B., Marcial, S., and Ambubuyog, G., this volume, Ground deformation prior to the 1991 eruptions of Mount Pinatubo.
Hadley, K.C., and Lahusen, R.G., 1991, Deployment of an acoustic flow-monitor system and examples of its application at Mount Pinatubo, Philippines: Eos, Transactions, American Geophysical Union, v. 72, no. 44, supplement, p. 67.
Hoblitt, R.P., Mori, J., and Power, J.A., this volume, Computer visualization of earthquake hypocenters.
Lee, W.H.K., ed., 1989, Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, 284 p.
Lee, W.H.K., Tottingham, D.M., and Ellis, J.O., 1989, Design and implementation of a PC-based seismic data acquisition, processing, and analysis system, in Lee, W.H.K., ed., Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, p. 21-46.
Lee, W.H.K., and Valdes, C.M., 1985, HYPO71PC: A personal computer version of the HYPO71 earthquake location program: U.S. Geological Survey Open-File Report 85-749, 30 p.
------1989, User manual for HYPO71PC, in Lee, W.H.K., ed., Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, p. 203-236.
March, G.D., and Murray, T.L., 1992, VOLPLOT; A PC-based program for viewing Cook Inlet Volcano-seismic data: U.S. Geological Survey Open-File Report 92-560-A, 6 p.
March, G.D., and Power, J.A., 1990, A networked computer configuration for seismic monitoring of volcanic eruptions: U.S. Geological Survey Open-File Report 90-422, 19 p.
Marso, J.N., and Murray, T.L., 1991, Real-time display of seismic spectral amplitude measurements: Examples from the 1991 eruption of Pinatubo volcano, Central Luzon, Philippines: Eos, Transactions, American Geophysical Union, v. 72, no. 44, supplement, p. 67.
Murray, T.L., 1990a, A user's guide to the PC-based time-series data-management and plotting program BOB: U.S. Geological Survey Open-File Report 90-56, 53 p.
------1990b, An installation guide to the PC-based time-series data management and plotting program BOB: U.S. Geological survey Open-File Report 90-634, 25 p.
------1992a, A low-data-rate digital telemetry system, in Ewert, J.W., and Swanson, D.A., eds., Monitoring volcanoes: Techniques and strategies used by the staff of the Cascades Volcano Observatory, 1980-90: U.S. Geological Survey Bulletin 1966, p. 11-24.
------1992b, A system for acquiring, storing, and analyzing low-frequency time-series data in near-real time, in Ewert, J.W., and Swanson, D.A., eds., Monitoring volcanoes: Techniques and strategies used by the staff of the Cascades Volcano Observatory, 1980-90: U.S. Geological Survey Bulletin 1966, p. 37-43.
Murray, T.L., Power, J.A., and Klein, F.W., 1993, PC_QPLOT, an IBM-PC compatible version of the earthquake plotting program QPLOT: U.S. Geological Survey Open-File Report 93-22, 17 p.
Parker, D.B., 1990, User manual for AcroSpin, in Lee, W.H.K., ed., Toolkit for plotting and displaying seismic and other data: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 2, p. 119-164.
Power, J.A., Murray, T.L., Marso, J.N., and Laguerta, E.P., this volume, Preliminary observations of seismicity at Mount Pinatubo by use of the Seismic Spectral Amplitude Measurement (SSAM) system, May 13-June 18, 1991.
Rogers, J.A., 1993, XDETECT version 3.18 user's reference guide: U.S. Geological Survey Open-File Report 93-261, 26 p.
Rogers, J.A., and Stephens, C.D., 1991, SSAM: a PC-based seismic spectral amplitude measurement system for volcano monitoring: Seismological Research Letters, v. 62, p. 22.
Swanson, D.A., Casadevall, T.J., Dzurisin, D., Malone, S.D., Newhall, C.G., and Weaver, C.S., 1983, Predicting eruptions at Mount St. Helens, June 1980 through December 1982: Science, v. 221, no. 4618, p. 1369-1376.
Tottingham, D.M., and Lee, W.H.K., 1989, User manual for XDETECT, in Lee, W.H.K., ed., Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, p. 89-118.
Tottingham, D.M., Lee, W.H.K., and Rogers, J.A., 1989, User manual for MDETECT, in Lee, W.H.K., ed., Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, p. 49-88.
Valdes, C.M., 1989, User manual for PCEQ, in Lee, W.H.K., ed., Toolkit for seismic data acquisition, processing and analysis: El Cerrito, Calif., Seismological Society of America, International Association of Seismology and Physics of the Earth's Interior Software Library, v. 1, p. 175-201.
Ward, P.L., 1989, SUDS: Seismic Unified Data System: U.S. Geological Survey Open-File Report 89-188, 123 p.
PHIVOLCS | University of Washington Press | U.S.Geological Survey
This page is <https://pubs.usgs.gov/pinatubo/murray/>
Contact: Chris Newhall
Last updated 06.11.99