FIRE and MUD Contents

A Comparison of Preeruption Real-Time Seismic Amplitude Measurements for Eruptions at Mount St. Helens, Redoubt Volcano, Mount Spurr, and Mount Pinatubo

By Elliot T. Endo,¹ Thomas L. Murray,¹ and John A. Power¹

¹U.S. Geological Survey.

ABSTRACT

Since 1985 we have had the opportunity to collect real-time seismic amplitude measurement (RSAM) data for preeruption periods at four different volcanoes. In this paper we introduce a technique to compare RSAM data corresponding to different magmatic eruptions. We normalized RSAM data and then used commercially available curve-fitting software for a personal computer to characterize RSAM data and simplify comparison. We found that the preeruption normalized RSAM data for three Mount St. Helens eruptions and one Mount Pinatubo dome-building eruption were best fit with an exponential equation of the type:

NORMALIZED_ RSAM_COUNT(t) = a+b exp (-t/c)

where NORMALIZED_RSAM_COUNT (t) corresponds to a normalized RSAM value at time t in decimal hours, and a, b, and c are parameters determined by the curve fitting program. While the exponential function is not the only type of equation to provide a satisfactory fit to some RSAM data, it is a convenient function to describe the similar patterns of increasing RSAM value and suggests that similar preeruption processes occurred in most of the dome-building eruptions we studied. Where good observational information was available, we know that dome-building eruptions followed periods of exponential increase in volcano-tectonic or "B" type earthquakes and a decline in seismic activity. We speculate that these effects are related to the migration of magma from depth to a level just below the surface, followed by less seismically active migration to the surface.

Results of curve fitting for normalized RSAM data for explosive eruptions at Redoubt Volcano, Mount Spurr, and Mount Pinatubo were mixed. Of five eruptions studied that resulted in explosive activity, three showed premonitory exponential-like increases in long-period earthquakes and brief periods of reduced or constant seismic activity shortly before the eruptions. The June 1992 eruption of Mount Spurr differed by having an explosive eruption during an exponential increase in volcano-tectonic seismicity. A high background seismic-noise level, numerous small, explosive eruptions prior to the June 15 paroxysmal eruption at Mount Pinatubo, and loss of some RSAM data precluded meaningful analysis and interpretation of RSAM data leading up to the paroxysmal eruption.

The andesitic dome-building eruption of Mount Pinatubo on June 7, 1991, and the explosive eruptions of volatile-rich andesite magma at Redoubt Volcano in December of 1989 and at Mount Spurr in 1992 have shorter durations for exponential-like increases in seismicity than do high-SiO₂ dome-building eruptions. Estimated durations for the rapid increases in seismicity at Redoubt Volcano, Mount Pinatubo, and Mount Spurr were typically less than 10 hours, or about one order of magnitude shorter in duration than dome-building eruptions at Mount St. Helens.

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.

INTRODUCTION

Real-time seismic amplitude measurement (RSAM) is a volcano-monitoring technique developed at the Cascades Volcano Observatory in Vancouver, Wash., that has been in use since 1985 (Endo and Murray, 1991). Since that time, preeruption RSAM data have been collected at Mount St. Helens, Redoubt Volcano, and Mount Spurr in the United States and at Mount Pinatubo in the Philippines. The RSAM technique has also been employed at the Mammoth Lakes area in California, at volcanoes on Hawaii, and in a number of countries in Central America and South America. The RSAM technique is a systematic electronic and computer method that provides a continuous measurement of average absolute seismic amplitudes for any number of seismic stations desired. Limitations of the technique are the number of seismic stations available for recording, electronics, and the computer hardware available. A potentially more serious limitation is that this simple technique does not discriminate between types of volcanic earthquakes, teleseismic events, regional earthquakes, wind, and other noise. Unlike the seismic spectral amplitude measurement (SSAM) technique, where user-defined spectral bands are monitored (Power and others, this volume; Stephens and others, 1994), all seismic signals are averaged and recorded. Briefly, the RSAM technique uses an analog-to-digital converter to convert analog seismic signals to a digital form suitable for computer storage and analysis. Sampled at a 50-Hz sampling rate, RSAM data are first averaged over 1 min and then averaged for a 10-min window for storage in a computer file. Each digital count represents 20 mV of analog seismic signal; thus, an average RSAM value is directly proportional to absolute average voltage of a seismic signal. The RSAM value is also proportional to the average ground velocity at the seismometer site. The digital form of the data and computer graphics provides a convenient method for near-real-time review of relative seismic activity. Details of the RSAM technique are described by Endo and Murray (1991) and Murray and Endo (1989).

Endo and others (1990) suggested that the migration of dacitic magma at Mount St. Helens was closely associated with the increase in RSAM counts, or the increase in the average amplitude of preeruption seismic signals (primarily volcano-tectonic events or "B" type earthquakes). Average ascent velocities were estimated for time periods defined by RSAM curves for two dome-building eruptions in 1986; however, there was no rigorous analysis of the RSAM data at that time. Assuming a linear relation and the approximate 24- to 48-h ascent times suggested by RSAM curves, Endo and others (1990) calculated a range of possible linear ascent velocities. Part of the problem in identifying the ascent time of magma was timing the onset of an eruption. Timed photography in October 1986 provided the best evidence for the onset of a dome-building eruption at Mount St. Helens. For that eruption there was a 12-hour interval between the first peak in RSAM data and the onset of the eruption. (Endo and others, 1990, reported a 6-h interval between a second peak in RSAM data and the onset of the eruption.) For other eruptions, the eruption onset was assumed to coincide with a substantial decrease in the rate of tilt close to the lava dome. Similar intervals, 12 to 18 h, between the peak in RSAM data and a rapid decrease in the rate of tilt adjacent to the lava dome (Endo and Murray, 1991) were observed for earlier eruptions in May 1986 and May 1985, respectively. The May 1986 dome-building eruption was preceded by almost 1 month of ash emissions and explosive activity.

The eruptions of Redoubt Volcano in 1989-90 (Power and others, 1994) provided the first opportunity after the October 1986 eruption of Mount St. Helens to evaluate RSAM data for another volcano. The swarm of long-period earthquakes (Chouet and others, 1994; Lahr and others, 1994; Stephens and others, 1994) that preceded the explosive eruption on December 14, 1989, produced an RSAM record that was at first difficult to analyze and interpret. A brief 6- to 8-h exponential-like increase starting December 13, 1991, at about 1030 local time was followed by about a 12-h linear increase in average RSAM amplitude. The linear increase was followed by a 5-h period of nearly constant amplitude, a 5-h period of decreasing average amplitude, and finally eruption of tephra (Power and others, 1994). This initial eruption was followed by many others during what Powers and other referred to as a "vent-clearing phase." This vent-clearing phase was followed by a dome-building phase that began about December 21, 1989. During the early part of this dome-building episode, volcano-tectonic events dominated. Starting around December 26, long-period earthquakes (Chouet and others, 1994) and tremor were dominant contributors to a rapid increase in RSAM amplitude counts. On January 2, 1990, this rapid increase leveled off a few hours before two large tephra eruptions at 1749 and 1927 local time (R. Page, written commun., 1993). The Redoubt RSAM data set from December 25, 1989, to January 2, 1990, differs from all other data sets of this report because it coincides with a period of known dome growth. However, owing to the rapid increase in cumulative RSAM counts, the data were used as a basis for issuing warnings before the eruptions on January 2.

RSAM data for volcano-tectonic earthquakes associated with a small dome-building eruption at Mount Pinatubo on June 7, 1991 (fig. 1), and RSAM data for a long-period earthquake swarm prior to an explosive eruption on June 14, 1991, had similarities to RSAM curves for eruptions at Mount St. Helens and Redoubt Volcano. Ewert and others (this volume), using tilt and seismic data, concluded that magma had been emplaced at shallow levels and possibly extruded on June 7, 1991. Visual observation on June 8 confirmed the presence of a new lava dome. The highest values in RSAM data for June 7 were probably related to a strong steam and ash emission (Ewert and others, this volume). RSAM data preceding the climactic eruption of Mount Pinatubo on June 15 are very complex, as a result of numerous smaller explosive eruptions starting on June 12 (Pinatubo Volcano Observatory Team, 1991; Hoblitt, Wolfe, and others, this volume) and did not lend themselves to detailed analysis and comparison to Mount St. Helens except for an increase in long-period seismicity (Power and others, this volume) prior to an explosive eruption at 1309 local time on June 14. The explosive events were part of a series that began on June 12 (Wolfe and Hoblitt, this volume).

Figure 1. A 6-min-wide section of the UBO seismogram (15.5 h long) showing the increase in volcano-tectonic earthquakes associated with the emplacement of magma at shallow depth beneath Mount Pinatubo on June 7, 1991. Corresponding RSAM counts are shown on the plot to the right. The time scale for the shaded area is not the same for the seismogram because of two changes in the translation rate for the seismogram.

The June 27, 1992, explosive eruption of Mount Spurr provided an example of RSAM data with a short but clear buildup in volcano-tectonic seismic activity. While not conspicuous, there was a discernible period of decrease in average seismic amplitudes about an hour prior to what was reported as the onset of the eruption (Power and others, in press). The August 18 eruption at Mount Spurr showed no significant precursory increase in RSAM counts of earthquake activity, and the September 17 eruption had 3 h of low-level tremor preceding the onset of the eruption activity and a few discrete events.

The purpose of this paper is to compare preeruption RSAM data from four stratovolcanoes and to determine whether there are any characteristics common to all preeruption RSAM data. This is a preliminary examination and is not intended as a comprehensive examination of all RSAM data for every eruptive episode for volcanoes such as Redoubt (Powers and others, 1994) or Pinatubo, which have undergone numerous eruptive episodes (Hoblitt, Wolfe, and others, this volume). We attempt to account for the apparent fine differences between peaks of seismic activity and onsets of dome-building eruptions at Mount St. Helens or initial vent-clearing eruption of Redoubt Volcano in December of 1989. Our definition of an eruption for this paper requires magma breaking through the surface of the volcano. That breakthrough could be in the form of a quiet dome-building eruption or a phreatomagmatic or magmatic explosive eruption.

We do not attempt to draw any conclusions regarding the eruption prediction value of RSAM for the paroxysmal eruption on June 15 because that is done by Cornelius and Voight (this volume).

RSAM DATA AND ANALYSIS TECHNIQUE

Data

The first preeruption RSAM data available for study were for the May 1985 dome-building eruption at Mount St. Helens. Two subsequent dome-building eruptions at Mount St. Helens in May 1986 and October 1986 provided additional data. For the May 1985 dome-building eruption we used 1-h-average RSAM data from the GDN (Garden) seismic station. For the 1986 dome-building eruptions we used 1-h-average RSAM data from the YEL (Yellow Rock) seismic station for analysis. RSAM data for October 1986 were corrected for a minus 6-dB gain change in the field at 1300 local time) on October 21. GDN was located about 900 m north of the geometric center of the lava dome at Mount St. Helens (fig. 2) and YEL 1,200 m north.

Figure 2. The Mount St. Helens area and the locations of seismic stations (black triangles). Contour interval 1,000 ft.

In 1989-90, the eruption at Redoubt Volcano in Alaska provided the first pre-eruption RSAM data for a second stratovolcano (Power and others, 1994). For this study, we used 10-min-average RSAM data leading up to the initial vent-clearing phases on December 14, 1989, and the dramatic January 2, 1990, eruption. Owing to the preliminary nature of this paper, data for about 20 other episodes from January to April 1990 (Stephens and others, 1994) were not examined. RSAM data from the RED seismic station (fig. 3) were used because the automatic gain-ranging seismic amplifier at seismic station RDN presented uncertainty in the relative amplitude of seismic signals, particularly when the amplifier reverted back to normal gain (Chouet and others, 1994). Seismic station RED was located approximately 7 km south of the active crater at Redoubt Volcano.

In 1991, the activity at Mount Pinatubo, Philippines, provided preeruption RSAM data for a third volcano. For this paper, preeruption 10-min-average RSAM data from seismic stations UBO and CAB (fig. 4) were compared with RSAM data from other volcanoes. Stations UBO and CAB were located at approximately 1 km east and 17 km east, respectively, from the old summit (Lockhart and others, this volume). While RSAM data were available for other seismic stations (PIE, BUR, GRN, PPO, and BUG), RSAM data for the small dome-building event accompanied by a strong ash emission on June 7, 1991, were not usable at most of the other seismic stations because of the poor signal-to-noise ratio. RSAM data from CAB on June 14 provided the best example for an increase in seismicity that preceded well timed explosive eruptions.

The most recent preeruption RSAM data available for comparison were from Mount Spurr, Alaska. During 1992, there were three eruptions at Mount Spurr (Power and others, in press; McNutt and others, in press). The first eruption, on June 27, 1992, produced a brief but gradual increase in RSAM counts. For the purpose of curve fitting, we used RSAM data that included the peak RSAM count, which was recorded about 3 h after the onset of the eruption (Power and others, in press). For Mount Spurr RSAM analysis, 10-min-average data were used for station BGL, located 7.5 km from the Crater Peak vent (fig. 5). RSAM data from CPK, located 400 m from the active vent, were not usable because of a gain-ranging problem with the amplifier.

The preeruption time window of RSAM data differed for each of the 10 eruptions studied. In each case, the starting time of a window was arbitrarily selected at some base level that corresponded to a period of relative seismic quiescence. For each of the Mount St. Helens eruptions and for two Mount Pinatubo eruptions, we looked at the preeruption data up to the time corresponding to the first peak in RSAM values. For other eruptions, we selected windows of data that appeared to have an exponential increase. For the June 27, 1992, eruption of Mount Spurr, we had to use RSAM data from the onset of the premonitory seismic swarm up to the peak in seismicity to have sufficient data for a fit. With the curve-fitting program it was important not to select points going into the period in which RSAM values were gradually decreasing. Table 1 provides the start and end times for each RSAM data set studied.

Figure 3. Redoubt Volcano and locations of the RED and RDN seismic stations (black triangles). (From Power and others, 1994). Contour interval 3,000 ft.

Figure 4. The Mount Pinatubo area showing locations of the UBO, CAB, and other seismic stations in operation before June 15, 1991.

Figure 5. Mount Spurr and locations of the BGL, CPK, and other seismic stations (black triangles). The shaded region is the approximate position of the Spurr caldera (from Power and others, in press).

Table 1. List of RSAM data start and end times for plots presented in this paper and minimum and maximum RSAM counts in each data set.

[Time is in Universal Time except for CAB and UBO, where local time was used. Minimum values are an estimated average in some cases. GDN and YEL data are for Mount St. Helens eruptions, RED for Redoubt Volcano, CAB and UBO for Mount Pinatubo, and BGL for Mount Spurr]

¹BGL RSAM counts above 1,200 may correspond to a clipped seismic signal.

Analysis

To facilitate comparison of RSAM data, we chose to normalize RSAM counts relative to their peak values. December 13-14, 1989, data for Redoubt Volcano were an exception to this procedure. For the December 13, 1989, RED RSAM data, we selected a time window that included only the initial 6 h of the swarm (Power, 1994) that preceded the eruption of December 14. After identifying a peak or maximum value, all points were divided by this value. Thus, the normalized peak or maximum value always had a value of 1. Absolute times for each of the RSAM average counts were converted to a simple decimal-hour scale starting at 0 for the first RSAM value selected for analysis.

To characterize RSAM data objectively we used the commercially available curve-fitting software Table Curve from Jandel Scientific. This IBM PC-compatible program uses automated statistical techniques to process data for the best curve-fitting equations. Version 3.0 of Table Curve uses 3,320 linear and nonlinear equations for its curve-fitting equation selection. Table Curve orders the valid equations, or equations that produce some measure of fit to the data, by the user selected goodness-of-fit criterion. For this paper the r² coefficient of determination is shown for equation ordering. Normalized RSAM data were analyzed for 9 of the 12 preeruption and eruption periods for which RSAM data were available.

CURVE-FITTING RESULTS

Mount St. Helens

Using its automatic search procedure, the program Table Curve produced equation-curve fits for three-parameter exponential equations (figs. 6, 7, and 8) as the highest ranking equations that provided fits for all three pre-eruption RSAM data sets available for Mount St. Helens' dome-building eruptions. Polynomial functions were the next most common type of equation that produced high r² coefficients of determination or high F-statistic values. While the fit of a polynomial function to the May 1985 RSAM data showed a comparable r² coefficient of determination in comparison to the exponential function, examination of residuals showed a weaker fit to RSAM data during the first 100 h of preeruption data. The parameters for the exponential equation fits and the corresponding r² coefficients of determination are given in table 2. The May 1986 and October 1986 dome-building eruptions yielded similar parameter values.

Figure 6. Preeruption RSAM data from the GDN seismic station for the May 1985 dome-building eruption at Mount St. Helens. The fitted exponential curve is shown as the solid line.

Figure 7. Preeruption RSAM data from the YEL seismic station for the May 1986 dome-building eruption at Mount St. Helens. The fitted exponential curve is shown as a solid line.

Figure 8. Preeruption RSAM data from the YEL seismic station for the October 1986 dome-building eruption at Mount St. Helens. The fitted exponential curve is shown as the solid line.

Table 2. Summary of parameters for the exponential fit of RSAM data for preeruption periods at Mount St. Helens (GDN, YEL) in 1985 and 1986, Redoubt Volcano (RED) in 1989-90, Mount Pinatubo (UBO, CAB) in 1991, and Mount Spurr (BGL) in 1992.

[As the plotted RSAM counts suggested by visual inspection, CAB RSAM data produced an unacceptable fit. a, b, and c are equation parameters determined by the program Table Curve. r² is the coefficient of determination. A value of 1.0 implies a perfect fit]

Redoubt Volcano

At first inspection, the December 13-15, 1989, RSAM data for Redoubt Volcano (RED seismic station) did not appear to have the potential for curve fitting like the Mount St. Helens data. A closer look at the data suggested a short exponential increase early in the long-period earthquake swarm that began on December 13. Table Curve selected a polynomial function for the highest ranked fit. For purposes of comparison to other RSAM data, we selected a fit for an exponential function. Results are shown in table 2 and figure 9. We refer to these author-selected fits as exponential-like increases. The RSAM counts for the period from December 25, 1989, to January 2, 1990, leading up to a series of explosive eruptions (Power and others, 1994) produced an exponential fit (fig. 10). This particular result differs from all others presented here because the RSAM data represent long-period seismicity associated with a period of dome growth. The result is presented without further comment.

Figure 9. Preeruption RSAM data from the RED seismic station at Redoubt Volcano for the early period of the long-period (LP) earthquake swarm that preceded the explosive eruption on December 14, 1989. The shaded inset shows the fitted exponential curve for the shaded time window.

Figure 10. Preeruption (eruption was on January 2, 1990) RSAM data from the RED seismic station at Redoubt Volcano. The fitted exponential curve is shown as a solid line.

Mount Pinatubo

For the small dome-building eruption on June 7, preeruption RSAM data for the UBO seismic station produced a three-parameter exponential-equation fit by use of Table Curve (fig. 11). The RSAM data associated with the June 7 dome-building eruption had similar maximum counts (or amplitudes) to the RSAM data examined for Redoubt Volcano. However, owing to seismic activity 5 km north-northwest of Mount Pinatubo, RSAM data were noisy and produced a relatively poor fit (r²<0.9). Initially viewed as complex, CAB RSAM data showed the possibility of an exponential increase for a 15-h window from June 13 to June 14 that did not appear to be contaminated by explosion earthquakes or large volcano-tectonic events. Table Curve selected a polynomial function for the highest rank fit. As for the early Redoubt Volcano RSAM data, we selected the exponential function for a subsequent fit. Those results are shown in figure 12 and table 2. We suspect that the high coefficient of fit (table 2) is an artifact of the small RSAM data set and that the parameters for the exponential fit do not reflect the true character of the increase in seismicity preceding the explosive eruptions. The peak in RSAM count on June 14 was followed by an explosive eruption about 4 h later (Power and others, this volume).

Preeruption fit to the June 11-15 CAB RSAM data leading up to the paroxysmal eruption on June 14 is complex and does not lend itself to simple analysis like the Mount Pinatubo data for June 7 and June 13-14. The r² coefficient of determination of about 0.7 (table 2) as well as residuals for the fit clearly indicate a poor fit. CAB RSAM data are shown in figure 12. However, the fitted curve is omitted. The Table Curve result is shown in table 2.

Figure 11. Preeruption (eruption was on June 7, 1991) RSAM data from the UBO seismic station at Mount Pinatubo. The highest RSAM values are probably associated with a strong gas and ash emission. The fitted exponential curve is shown as a solid line.

Figure 12. A 20-h window of preeruption normalized RSAM data from the CAB seismic station and fitted curve for the period preceding the first explosive eruption on June 14, 1991. The corresponding time window is shown as a shaded rectangle on the 0- to 350-h plot of CAB RSAM data. The fitted curve for the 0- to 350-h CAB RSAM data has been deleted. All times are local time.

Mount Spurr

Mount Spurr provided some of the least noisy RSAM data available to date. RSAM data from the BGL station for the period before and during the June 1992 eruption were best fit by an exponential curve and gave the best r² coefficient of determination for all of the RSAM data analyzed (fig. 13, table 2). Table Curve also found acceptable fits to dozens of other equations that have curves similar to the exponential equation. Subsequent eruptions in August and September showed no similar gradual increase in seismicity at BGL. However, the September eruption showed a brief increase on the CPK and the CRP records (McNutt, written commun., 1993). Onsets of the latter two eruptions were relatively sudden (fig. 14) on the BGL RSAM seismic record.

Figure 13. Preeruption and eruption-related RSAM (June 1992) data from the BGL seismic station for Mount Spurr. The fitted exponential curve is shown as a solid line.

Figure 14. A, Graph of RSAM data from BGL seismic station, Mount Spurr, spanning the period of the June 1992 eruption. The rapid increase in RSAM counts, or average seismic amplitude, took place over a period of approximately 7 to 8 h. B, Graph of RSAM data from BGL seismic station spanning the period of the August 1992 eruption. No preeruption increase was observed for this eruption. C, Graph of RSAM data from BGL seismic station spanning the period of the September 1992 eruption. Low-level tremor prior to the August and September eruptions had almost no impact on RSAM counts.

THE THREE-PARAMETER EXPONENTIAL EQUATION

The equation ranked number one by the r² coefficient of determination or F-statistic for all cases preceding dome-building eruptions was the exponential equation:

NORMALIZED_ RSAM_COUNT(t) = a+b exp (-t/c)

where NORMALIZED_RSAM_COUNT(t) corresponds to an RSAM value at time t in decimal hours for the RSAM value. a, b, and c are unknown parameters determined by Table Curve. For the purpose of comparison, the exponential equation was selected for fitting RSAM data or seismicity associated with explosive eruptions where the exponential function was not selected automatically by Table Curve.

The a parameter corresponds to the initial base level, or vertical offset, at 0 time. This parameter is related to the ratio of RSAM counts to the maximum RSAM value. A low background noise level or low counts and a large number of counts for the peak RSAM value result in a relatively small a parameter. RED seismic station for Redoubt and UBO for Mount Pinatubo had the lowest ratio for peak signal to background noise level and, hence relatively higher a values (tables 1 and 2). Seismic stations at Mount St. Helens commonly had the highest ratio of maximum RSAM counts to background noise level, along with BGL at Spurr volcano, hence, low a values. High a values appear to have a negative impact on how well RSAM data define a particular type of function. An extreme case would be an a value almost equal to 1.0. That would indicate a signal-to-noise ratio of 1.0.

The b parameter is a scale factor related to the normalized RSAM data. If RSAM data were not normalized, this would be related to seismic station gain, proximity to the seismic source, background seismic noise level, seismic station site response, and seismic station instrumentation response. We have normalized RSAM data to allow meaningful comparison of RSAM data from different volcanoes.

Parameter c is related to the slope, or curvature, of the exponential curve and the time window for the RSAM data. Thus, a large absolute value indicates a relatively long time window, or run-up time, for RSAM data and relatively slow changes in RSAM counts from an early stage. Station RED for the period leading up to the eruption on January 2, 1990, is such an example. The time interval from a reference level to a peak value was about 220 h. A small c parameter indicates a short time window and a rapid increase in RSAM counts. The BGL RSAM data for Mount Spurr produced the smallest c parameter and one of the shortest run-up times. BGL RSAM data went from a base reference level to a peak value in less than 7 h. Mount St. Helens preeruption RSAM data for the two dome-building eruptions in 1986 produced similar c parameter values and run-up times. The c value determined for UBO RSAM data for the June 7, 1991, dome-building eruption at Mount Pinatubo differs by about a factor of two less. Results of all curve fits completed are shown in table 2.

The three-parameter exponential equation was not the only equation type that produced acceptable fits to the RSAM data selected for this study. The fact that it was a common equation type that fit a number of preeruption RSAM data does suggest an exponential component to the increase in average seismic amplitude and provides a basis for comparison of RSAM data. The three-parameter exponential equation provides a simple method for comparing RSAM data from different volcanoes and seems appropriate for the purpose of this paper because the exponential function is frequently used to describe natural processes such as growth of a forest, growth of the population of the Earth, and radioactive decay (Gellert and others, 1975).

DISCUSSION

While the increase in RSAM counts appears to have some value in forecasting or prediction of eruptions (Voight and Cornelius, 1991; Cornelius and Voight, 1994; Cornelius and Voight, this volume), the application of curve-fitting techniques to predicting precise times of eruptions needs to be viewed with caution. The process of magma migration to shallow levels in the crust and eruptions is complex. From detailed studies of drill core from the Inyo Domes, California, Westrich and others (1988) concluded that, at least for rhyolitic magma, there was a significant net increase in magma viscosity during the final 1 km of magma ascent owing to water exsolution and bubble growth that result from decompression. This change in bulk viscosity apparently was the reason that much of the Inyo dike failed to reach the surface. Besides having an impact on the ascent rate of magma, water exsolution and bubble growth have an impact on whether an intrusion results in explosive activity, relatively quiet dome building, or extrusion as a lava flow (Eichelberger and others, 1986). Westrich and others (1988) state, "The rate and extent to which exsolved water can escape the system control whether or not the magma fragments to form tephra or extrudes intact as lava."

The first explosive eruption on December 14, 1989, at Redoubt Volcano took place about 10 h after the peak in long-period earthquake RSAM counts and about 23 h after the onset of the swarm. Lahr and others (1994) reported hypocenter depths of 1.4 km below the crater floor for this group of long-period earthquakes. We suspect that the long-period swarm was a direct result of water exsolution and bubble growth. Chouet and others (1994) suggest interaction of magma with ground water. The explosive eruption at 1309 local time on June 14 at Mount Pinatubo was similarly preceded by an earthquake swarm dominated by long-period earthquakes. The explosive eruption took place about 4 h after a peak in RSAM counts. While weak, there is a suggestion of a decrease in volcano-tectonic seismic activity about 1 h prior to what has been called the onset of the eruption at Mount Spurr (Power and others, in press; McNutt and other, in press). For volatile-rich andesitic or dacitic magma there is a catastrophic disruption of the magma during magma ascent or following shallow emplacement because of substantial overpressures from volatiles and the onset of fragmentation at the upper levels in the conduit.

Dome-building eruptions like those at Mount St. Helens in 1985 and 1986 involve significantly degassed magma and apparently represent a low ratio of volatile exsolution to volatile escape from the system. In contrast, explosive eruptions like those of Redoubt Volcano in 1990 and Mount Pinatubo in 1991 represent a volatile-rich magma and a high ratio of volatile exsolution to volatile escape from the system between explosive episodes. The work of Westrich and others (1988) suggests that, for either case, magma ascent is slowed because of volatile exsolution and bubble growth resulting from decompression, particularly at depths above 1 km. That slowing down should not be confused with the approximately 12-h interval between the first peak in RSAM counts for the October 1986 dome-building eruption at Mount St. Helens and the actual extrusion that was confirmed by timed photography. For Mount St. Helens, that time interval of less seismic activity represents the time required for magma to ascend through the relatively weak pyroclastic material and the lava flows of the volcanic edifices.

Our results do not suggest that all eruptions have to be preceded by exponential or exponential-like increases in volcano-tectonic or long-period seismicity. Numerous explosive gas and ash emissions at Mount St. Helens in April and May of 1986 were not preceded by precursory seismic activity. Following the last dome-building eruptions in 1986, numerous ash emissions and explosions (Mastin, 1994) at Mount St. Helens typically did not show premonitory increases in seismic activity and RSAM counts. These eruptions did not involve the ascent of magma. Mastin (1994) hypothesized that "the explosion-like seismic events represent the transport of gas, probably of magmatic origin, in the shallow subsurface." Not all explosive eruptions at Redoubt Volcano were preceded by precursory long-period activity (Stephens and others, 1994). At Mount Pinatubo, several explosive eruptions before June 14 were preceded by significantly less seismicity (Hoblitt, Wolfe, and others, this volume). The August 1992 explosive eruption of Mount Spurr showed no signs of detectable earthquakes prior to the eruption, and the September eruption was preceded by some discrete events a few hours before the eruption. Episodes of tremor preceded the June, August, and September 1992 eruptions (AVO, 1993). During the 1988-89 eruptive activity at Mount Tokachi, Japan, many explosive eruptions did not show any detectable precursory earthquake activity (Nishimura and others, 1990). Only one-third of 23 observed explosive eruptions were preceded by small "low frequency" earthquakes (Okada and others, 1990). Eruption types, associated magma or lava types, dominant earthquake types, and approximate period of increased seismicity or an exponential-like buildup for each eruption are shown in table 3.

The primary intention of this paper was to investigate the possibility of the existence of common characteristics in the increase of preeruption RSAM data and precursory seismicity. We found that for most dome-building preeruption RSAM data, it was possible to find a function or equation type that fit the gradual increase in RSAM counts prior to an eruption. While it may appear remarkable that the exponential equation was the one ranked highest in all cases, the exponential equation was not the only type that produced acceptable fits to the RSAM data. However, it was the equation type that most commonly produced an acceptable fit (high r² coefficient) to most of the RSAM data and was the most convenient for comparing increases in RSAM counts.

The similarity of fits suggests that a common physical process is in operation at two volcanoes (Mount St. Helens and possibly Mount Pinatubo). At Mount St. Helens, we speculated (Endo and others, 1990) that the relative average seismic amplitude at a seismic station was directly related to the ascent of magma from a depth of about 1.4 km to just below the surface. If that is the case, then the exponential equation that fits the RSAM data analyzed for this paper is related to a magma velocity function. If the depth of origin for the magma is known, then a full expression for magma ascent velocity can be determined, and the normalized RSAM plot could be similar to a plot of time versus distance traveled. If the exponential increase is related to ascent of magma, the implication is that the initial ascent velocities are small and increase substantially over time. Unfortunately, the Mount St. Helens RSAM data suggest that the relation is far more complex. While the exponential equation parameters are nearly identical for the fit of normalized RSAM data for eruptions of similar volume in 1986 (Swanson and Holcomb, 1989), they differ for an eruption of smaller volume in 1985.Assuming no large differences in the conduit system, a possible explanation is that the smaller volume of eruptible magma ascended at a relatively slower velocity.

Redoubt Volcano, Mount Pinatubo (June 13-14), and Mount Spurr RSAM data were more difficult to analyze and interpret. In some cases the exponential equation was not selected automatically by the curve-fitting program. It is likely that the increases in RSAM data for the different eruptions studied are more complex. Furthermore, exponential increases in RSAM data for long-period earthquakes are probably not directly related to magma ascent. While the study is far from conclusive, there is a suggestion that the increases in seismicity (long-period or volcano-tectonic earthquakes) prior to explosive andesitic and dacitic eruptions are an order of magnitude shorter in duration (run-up time) than increases in seismicity associated with dacitic dome-building eruptions (table 3).

Table 3. Volcanoes, eruption types, magma or lava types, dominant earthquake type, and approximate durations (run-up time) of rapid increase in precursory seismicity.

[MSH, Mount St. Helens; RV, Redoubt Volcano; MP, Mount Pinatubo; MS, Mount Spurr. Tokachi-dake Volcano precursory times were estimates from an amplitude-variation figure in Ikeda and others (1990). The Unzen Volcano precursory time is from Shimizu and others (1992). Low-frequency earthquakes at Tokachi-dake Volcano and Unzen Volcano may be "B" type rather than true long-period earthquakes described by Chouet and others (1994). VT, volcano-tectonic]

¹Swanson and Holcomb, 1990.

²Nye and others, 1994.

³The January 2, 1990, explosive eruption was preceded by dome building.

⁴Pallister and others, this volume.

⁵Harbin and others, in press.

⁶Nye and others, in press.

⁷Ikeda and others, 1990.

⁸Yanagi and others, 1992.

CONCLUSION

The main results of our comparison of preeruption seismicity by use of RSAM data follow.

1. The rapid increase in seismicity prior to dacitic and andesitic dome-building eruptions can be characterized by an exponential function.
2. For dacitic (Mount St. Helens) and andesitic (Mount Pinatubo, June 7, 1991) dome- building eruptions, we speculate that the exponential increase corresponds to the migration of magma from shallow depths (approximately 1.4 km at Mount St. Helens) to just below the surface. The exponential function is proportional to ascent velocity for magma. For Mount St. Helens, the time interval between the peak in RSAM counts and extrusion is believed to be the time required for dacitic magma to migrate through the relatively weak pyroclastic deposits of the edifice.
3. Not all andesitic eruptions are preceded by an increase in seismicity. For this case we speculate that magma migrates in well-established conduits with relatively few  detectable earthquakes. Rapid exponential-like increases may precede the initial eruption or may occur in cases where the conduit is blocked.
4. Detectable long-period, volcano-tectonic, or "B" type earthquakes or tremor are not essential precursors to explosive gas and ash emissions.
5. Eruptions, dome-building or explosive, are sometimes preceded by a decrease in seismicity or a leveling off in seismicity an hour or longer before extrusions or explosive eruptions. For dome-building eruptions, that decrease, as at Mount St. Helens, may be related to the migration of magma into the volcano edifice. For explosive eruptions, the decrease may represent a transient equilibrium between volatile pressure and lithostatic pressure.

We believe accurate characterization of precursory seismic activity by using the RSAM monitoring technique is just one step forward in the understanding of the volcanic eruption process. Petrologic studies, gas-chemistry studies, experimental studies of bubble growth, tilt and other earth-deformation studies, studies of seismic source parameters, and so on, are all required to relate the increase in seismic activity better, or, in some cases, the absence of increased seismicity to volcanic eruptions so we can advance our understanding of the magma migration and eruption process.

REFERENCES CITED

Alaska Volcano Observatory, 1993, Mount Spurr's 1992 eruptions: Eos, Transactions, American Geophysical Union, v. 74, p. 217-222.

Brantley, S., ed., 1990,The eruption of Redoubt Volcano, Alaska, December 14, 1989-August 31, 1990: U.S. Geological Survey Circular 1061, 33 p.

Chouet, B.A., Page, R.A., Stephens, C.D., Lahr, J.C., and Power, J.A., 1994, Precursory swarms of long-period events at Redoubt Volcano (1989-1990), Alaska: Their origin and use as a forecasting tool, in Miller, T.P., and Chouet, B.A., eds., The 1989-1990 eruptions of Redoubt Volcano, Alaska: Journal of Volcanology and Geothermal Research, v. 62, p. 95-135.

Cornelius, R.R., and Voight, B., 1994, Seismological aspects of the 1989-1990 eruption at Redoubt Volcano, Alaska: The materials failure forecasting method (FFM) with RSAM seismic data: Journal of Volcanology and Geothermal Research, v. 62, p. 469-498.

Cornelius, R.R., and Voight, B., this volume, Real-time seismic amplitude measurement (RSAM) and Seismic spectral amplitude measurement (SSAM) analyses with the Materials Failure Forecast Method (FFM), June 1991 explosive eruption at Mount Pinatubo.

Eichelberger, J.C., Carrigan, C.R., Westrich, H.R., and Price, R.H., 1986, Non-explosive silicic volcanism: Nature, v. 323, p. 598-602.

Endo, E.T., Dzurisin, D., and Swanson, D.A., 1990, Geophysical and observational constraints for ascent rates of dacitic magma at Mount St Helens, in Ryan, M.P., ed., Magma transport and storage: John Wiley & Sons, p. 318-334.

Endo, E.T., and Murray, T., 1991, Real-time seismic amplitude measurement (RSAM): A volcano monitoring 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.

Gellert, W., Gottwald, S., Hellwich, M., Kastner, H., Kustner, H., ed., 1975, The VNR concise encyclopedia of mathematics: New York, Von Nostrand Reinhold, 776 p.

Harbin, M.L., Swanson, S.E., Nye, C.J., and Miller, T.P., in press, Preliminary petrology and chemistry of proximal eruptive products: 1992 eruptions of Crater Peak, Mount Spurr, Alaska: U.S. Geological Survey Bulletin.

Hoblitt, R.P., Wolfe, E.W., Scott, W.E., Couchman, M.R., Pallister, J.S., and Javier, D., this volume, The preclimactic eruptions of Mount Pinatubo, June 1991.

Ikeda, Y., Katsui, Y., Nakagawa, M., Kawachi, S., Watanabe, T., Fujibayashi, N., Shibata, T., and Kagami, H., 1990, Petrology of the 1988-1989 essential ejecta and associated glassy rocks of Tokachi-dake volcano in central Hokkaido, Japan: Bulletin of the Volcanological Society of Japan, v. 35, no. 2, p. 47-163.

Lahr, J.C., Chouet, B.A., Stephens, C.D., Power, J.A., and Page, R.A., 1994, Earthquake classification, location, and error analysis in a volcanic environment: Implications for the magmatic system of the 1989-1990 eruptions at Redoubt Volcano, Alaska: Journal of Volcanology and Geothermal Research, v. 62, p. 137-151.

Lockhart, A.B., Marcial, S., Ambubuyog, G., Laguerta, E.P., and Power, J.A., this volume, Installation, operation, and technical specifications of the first Mount Pinatubo telemetered seismic network.

Mastin, L.G., 1994, Explosive tephra emissions at Mount St. Helens, 1989-1991: The violent escape of magmatic gas following storms: Geological Society of America Bulletin, v. 106, p. 175-185.

McNutt, S., Tytgat, G., and Power, J.A., in press, Preliminary analysis of volcanic tremor associated with the 1992 eruptions of Crater Peak, Mount Spurr, Alaska: U.S. Geological Survey Bulletin.

Murray, T.L., and Endo, E.T., 1989, A real-time seismic amplitude measurement system (RSAM): U.S. Geological Survey Open-File Report 89-684, 21 p.

Nishimura, Y., Miyamachi, H., Ueki, S., Nishimura, T., Shimizu, H., Ohmi, S., and Okada, H., 1990, Joint seismometrical observations by the National University Team during the 1988-1989 eruptive activity of Mt. Tokachi, Hokkaido: Bulletin of the Volcanological Society of Japan, v. 35, no. 2, p. 163-172.

Nye, C.J., Harbin, M.L., Miller, T.P., Swanson, S.E., Neal, C.A., in press, Whole-rock major- and trace-element chemistry of 1992 ejecta from Crater Peak, Mount Spurr, Alaska: A preliminary overview: U.S. Geological Survey Bulletin.

Nye, C.J., Swanson, S.E., Avery, V.F., and Miller, T.P., 1994, Geochemistry of the 1989-1990 eruption of Redoubt Volcano, pt. I: Whole-rock major and trace element chemistry: Journal of Volcanology and Geothermal Research, v. 62, p. 429-452.

Okada, H., Nishimura, Y., Miyamachi, H., Mori, H., and Ishihara, K., 1990, Geophysical significance of the 1988-1989 explosive eruptions of Mt. Tokachi, Hokkaido, Japan: Bulletin of the Volcanological Society of Japan, v. 35, no. 2, p. 176-203.

Pallister, J.S., Hoblitt, R.P., Meeker, G.P., Knight, R.J., and Siems, D.F., this volume, Magma mixing at Mount Pinatubo: Petrographic and chemical evidence from the 1991 deposits.

Pinatubo Volcano Observatory Team, 1991, Lessons from a major eruption: Mount. Pinatubo, Philippines: Eos, Transactions, American Geophysical Union, v. 72, 4 p.

Power, J.A., Jolly, A.D., Page, R.A., and McNutt, S.R., in press, Seismicity and forecasting of the 1992 eruptions of Crater Peak vent, Mount Spurr, Alaska: An overview: U.S. Geological Survey Bulletin.

Power, J.A., Lahr, J.C., Page, R.A., Chouet, B.A., Stephens, C.D., Harlow, D.H., Murray, T.L., and Davies, J.N., 1994, Seismic evolution of the 1989-1990 eruption sequence of Redoubt volcano, Alaska: Journal of Volcanology and Geothermal Research, v. 62, p. 153-182.

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.

Shimizu, H., Umakoshi, K., Matsuwo, N., and Ohta, K., 1992, Seismological observations of Unzen Volcano before and during the 1990-1992 eruption, in Yanagi, T., Okada, H., and Ohta, K., eds., Unzen Volcano the 1990-1992 eruption: Fukuoka, Japan, The Nishinippon and Kyushu University Press, p. 38-43.

Stephens, C.D., Chouet, B.A., Page, R.A., and Lahr, J.C., in press, Seismological aspects of the 1989-1990 eruptions at Redoubt Volcano, Alaska: The SSAM perspective: Journal of Volcanology and Geothermal Research.

Swanson, D.A., and Holcomb, R.T., 1989, Regularities in the growth of the Mount St. Helens dacite dome, 1980-1986, in Fink, J.H., ed., Lava flows and domes, emplacement mechanisms and hazards: Berlin, Springer, 34 p.

Voight, B., and Cornelius, R.R., 1991, Prospects for eruption prediction in near real-time: Nature, v. 350, p. 695-698.

Westrich, H.R., Stockman, H.W., and Eichelberger, J.C., 1988, Degassing of rhyolitic magma during ascent and emplacement: Journal of Geophysical Research, v. 93, p. 6503-6511.

Wolfe, E.W. and Hoblitt, R.P., this volume, Overview of the eruptions.

Yanagi, T., Nakada, S., and Maeda, S., 1992, Temporal variation in chemical composition of the lava extruded from the Jigokuato Crater, Unzen Volcano, in Yanagi, T., Okada, H., and Ohta, K., eds., Unzen Volcano the 1990-1992 eruption: Fukuoka, Japan, The Nishinippon and Kyushu University Press, p. 82-91.

FIRE and MUD Contents

PHIVOLCS | University of Washington Press | U.S.Geological Survey
This page is <https://pubs.usgs.gov/pinatubo/endo/>
Contact: Chris Newhall
Last updated 06.10.99