FIRE and MUD Contents
1U.S Geological Survey.
2Philippine Institute of Volcanology and Seismology.
The preclimactic eruptive activity at Mount Pinatubo in late May and early June of 1991 consisted of dome growth, followed by 4 vertical eruptions, followed by 13 surge-producing eruptions. The dome and pyroclasts from the first vertical eruption are composed almost entirely of andesite. In subsequent eruptions the proportion of andesite pyroclasts declined irregularly as the proportion of dacite pyroclasts increased; pyroclasts in the late surges are composed almost entirely of dacite. That pyroclasts from most preclimactic deposits display a wide density range suggests that the early vanguard magmas were subject to substantial degassing before eruption.
Stratigraphic evidence indicates that eruption magnitude declined successively through the vertical eruptions and generally declined successively through the surge-producing eruptions. This decline in magnitude was accompanied by a general decline in the repose-interval duration, so the successive eruptions were generally smaller and more closely spaced in time. The duration of successive eruptions at first declined but then apparently increased somewhat just before the climactic eruption. This behavior is mimicked by a model wherein the pressurization rate of magma at the top of the magma conduit increases from a value initially inadequate to sustain a continuous eruption and wherein the confining pressure, controlled by degassing at the top of the magma column during repose intervals, decreases with time. The climactic eruption began when the pressurization rate, controlled by the influx of magma from a deep reservoir, exceeded the rate of depressurization, as magma was expelled during eruptions.
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.
One of the greatest challenges in volcano-hazard-mitigation efforts is to forecast eruptive behavior correctly. Because mitigation measures depend on anticipation of eruption types, timing, and the areas likely to be affected, accurate forecasts can minimize the social, political, and economic disruption caused by eruptions. For example, vigorous, convectively driven, vertical eruption plumes disperse tephra over large regions, whereas the effects of pyroclastic density currents are usually far more restricted and far more devastating. Areas subject to pyroclastic density currents need to be evacuated; tephra-prone areas need not be evacuated, but a variety of other mitigative measures should be implemented.
Eruptive behavior remains difficult to forecast with confidence. Forecasts are usually based on data gleaned from deposits produced by past eruptions, from the eruptive behavior of other similar volcanoes, and from historical activity at the volcano under scrutiny. Stratigraphic data are often incomplete, and behavior at different volcanoes or even the same volcano can vary within wide limits. Compare, for example, the Pinatubo eruptive sequence to two other recent sequences: the eruptions of Mount Lamington, Papua New Guinea, in 1951 (Taylor, 1958) and El Chichón volcano, Chiapas, Mexico, in 1982 (Sigurdsson and others, 1984). All three sequences occurred at volcanoes that had been in repose for hundreds of years, all three produced vigorous convectively-driven vertical eruption plumes as well as dilute pyroclastic density currents (surges), and all were driven by sulfur-rich andesite or dacite magmas.
The Pinatubo sequence, to be described in detail below, consisted sequentially of increasing ash emissions, dome growth, 4 vertical eruptions with continued dome growth, 13 pyroclastic surges, and a climactic vertical eruption with associated pyroclastic flows. Postclimactic activity (through 1993) consisted of declining ash emissions and emplacement of a small dome. The El Chichón sequence consisted of an initial major vertical eruption followed 5 days later by two more major eruptions. The second major eruption began with a surge and then progressed into a vertical eruption that also culminated with a surge. The third major eruption began as a vertical eruption and culminated with a minor surge event. Following a week of increasingly vigorous ash emission, Mount Lamington produced a major pyroclastic surge; a large vertical eruption occurred later the same day. Vertical eruptions continued sporadically over the next 1 months and were succeeded by dome growth.
Despite some similarities, none of the sequences could serve as a reliable predictor of the others. One route to better eruption-pattern forecasts may be through a better understanding of the fundamental processes that control eruptive behavior. To this end, we undertook a reconnaissance study of the deposits produced by Pinatubo's preclimactic eruptions, in the hope that the deposits might provide some insights into the controlling processes. We present here a summary of the preclimactic behavior along with granulometry, component analyses, and densiometry of deposits, as well as some other relevant observations, and discuss these data in terms of processes that may control eruptive behavior.
In the following discussion, we refer to two basic types of pyroclastic transport processes: pyroclastic fall and pyroclastic density current. We use these terms and their variants in the following way.
Pyroclastic fall.--A rain of pyroclasts that fall to earth after vertical transport within an eruption plume. Particle concentration is low, and interactions between falling particles or particle aggregates are insignificant. The resulting deposit has wide areal extent and is typically well sorted; deposit thickness shows little variation locally, except on slopes that exceed the angle of repose. Topography exerts little or no influence on the areal distribution.
Pyroclastic density current.--A mixture of pyroclasts and gases whose net density is greater than that of the surrounding atmosphere. The mixture flows en masse along the ground like a fluid. We distinguish two types of pyroclastic density current. A pyroclastic flow is a dense, high-concentration gas-pyroclast mixture that typically produces massive, poorly sorted deposits that pond in topographically low areas. A pyroclastic surge is a dilute suspension of pyroclasts in turbulent gas. Particle concentrations are transitional between pyroclastic flow and fall, and deposits are typically relatively thin, show low-angle crossbedding, dune structures, and exhibit topographic control and sorting values intermediate between those of pyroclastic-flow and pyroclastic-fall deposits. We use the modifier "slack" to refer to surges whose densities are just marginally more dense than the surrounding air; these typically produce thin, parallel beds of well-sorted fine ash.
Pyroclastic-surge deposits emplaced on June 14 and 15 within 10 km of Pinatubo's pre-June 15 summit are termed "proximal"; those emplaced at greater distances are termed "distal."
We use the term "pyroclastic fountain" to denote the vertical ejection of a mixture of gas, ash, and larger pyroclasts that falls back to earth, analogous to a vertical water fountain (Hoblitt, 1986, fig. 16C). This phenomenon has been termed "column collapse" (Wilson and others, 1980).
Pinatubo's preeruption summit provides a convenient reference point for distance estimates. We refer to this point as the "old summit."
It is useful to divide the 1991-92 volcanic activity at Mount Pinatubo (fig. 1) into a series of eight behavioral phases (Wolfe and Hoblitt, this volume). Most behavioral changes were abrupt, such as the onset of vertical eruptions on June 12. The change from Phase I to Phase II, however, was gradual, and the boundary differed for different phenomena: ash emissions began to increase in late May, SO2 emissions reached a peak on May 28 (Daag, Tubianosa, and others, this volume), and seismicity began to escalate on June 1. For convenience, we have chosen June 1 as the boundary between Phase I and Phase II.
We describe briefly Phase I, then focus on Phases II, III, IV, and V; a summary of the narrative is contained in table 1. Eruption durations for Phases IV and V were estimated from seismograms and Seismic Spectral Amplitude Measurement (fig. 2A,B; SSAM), which monitors the spectral content of earthquakes (Rogers, 1989; Power and others, this volume). Eruption onset times were determined from seismograms, microbarograph records (fig. 2C), and direct visual observations. Plume-height estimates in table 1 are based on C-band radar observations, primarily from Clark Air Base (Clark; fig. 1), 21 km east of Mount Pinatubo. The plume heights for the first three eruptions are minimum estimates. During the first eruption, Clark radar operators were evacuated from their station while the plume was still rising; the plumes of the second and third eruptions exceeded the radar height limit of 24 km. The first eruption almost certainly also exceeded this limit.
1991 Eruption Phases
March 15-May 31
Felt earthquakes beginning on March 15, phreatic explosions on April 2, continuing emission of steam and minor ash, constant release of seismic energy, and generally increasing SO2 emission.
Escalating seismic-energy release, concentration of shallow earthquake hypocenters beneath Pinatubo's summit, diminution in SO2 emission, inflationary tilt, and increasing ash emission culminating in the birth of a lava dome on June 7.
Dome growth, increasingly heavy emission of ash, escalating seismic-energy release.
Four vertical eruptions with minor pyroclastic flows, continued dome growth, and heavy emission of ash.
Multiple surge-producing eruptions beginning at about 1516 on June 14 and continuing until the climactic phase.
The climactic phase of activity, beginning at 1342, consisting of a large vertical eruption with the production of voluminous pumiceous pyroclastic flows. This phase culminated in formation of a collapse caldera.
Postclimactic phase of activity wherein initially voluminous, continuous ash emissions slowly waned through the middle of late July; intermittent small ash eruptions continued until early September.
1992 Eruption Phase
Growth of small dome of andesite lava.
Figure 1. Major landmarks in the vicinity of Mount Pinatubo.
Figure 2. A, SSAM record (0.5- to 1.5-Hz band, station Clark) from 1200 on June 11 to 0000 on June 14, 1991; blue bands indicate eruptions. B, Same as A, from 0000 on June 14 to 1200 on June 16. C, Barogram recorded at Clark from about 0200 on June 14 to about 1200 on June 16; blue bands indicate eruptions. Each surge-producing eruption corresponds to a barogram deflection. The deflection corresponding to the 0257 June 15 eruption precedes that eruption by about 30 min; the discrepancy is absent on the Cubi Point barogram (not shown). This suggests that the pen on the Clark barograph malfunctioned temporarily.
Table 1. Chronology of eruptive activity at Mount Pinatubo during June and July 1991.
[Explosive-eruption onset times determined from seismograms supplemented by microbarograph records and, as noted, from direct visual observations. We estimate +-3 min uncertainty in absolute time. Seismic durations determined from seismograms and 1-min-averaged data from Seismic Spectral-Amplitude Measurements (SSAM) system in the 0.5- to 1.5-Hz band (fig. 2A, B; SSAM observations are described by Power and others, this volume). Events prior to 0921 June 12 determined from seismic records of station PIE (9 km northeast of Mount Pinatubo summit; events later than 0921 June 12 determined from seismic records of station CAB (17 km northeast of summit). Plume heights from visual and military weather-radar observations. LP, long-period]
On April 2, following some initial seismic activity, small explosions occurred along a 1.5-km-long northeast-trending fracture across Mount Pinatubo's northeast flank. Shortly after the explosions, an aerial inspection revealed nine vigorous fumaroles across the volcano's upper north flank, southwest of the fracture (Sabit and others, this volume). Three of the fumaroles (Wolfe and Hoblitt, this volume) soon became dominant, emitting copious quantities of steam and a minor amount of ash. The estimated SO2 flux increased from 500 t/d on May 13, the day of the first measurement, to 5,000 t/d on May 28 (Daag, Tubianosa, and others, this volume).
In late May, fumarole emissions, which had been almost entirely steam, began to contain an increasing proportion of ash (fig. 3). This change was gradual and variable; periods of increased ash emission alternated with periods of steam emission. Prevailing winds carried ash over the South China Sea, about 33 km to the west. By May 30 we were able to collect ash adhering to vegetation a few kilometers west of the volcano. On May 28, SO2 emissions reached a peak of 5,000 t/d and then decreased to a minimum of a few hundred tons per day on June 5; on June 7 the rate was about 1,000 t/d (Daag, Tubianosa, and others, this volume).
Seismicity began to change in early June. Previously, most earthquakes were located about 5 km northwest of Pinatubo's summit, at depths of about 3 to 7 km. A lesser number were located just north of the summit, beneath the fumaroles, at a depth of about 1 to 3 km (Harlow and others, this volume). Episodes of low-amplitude volcanic tremor occurred sporadically; RSAM values--used to monitor the rate at which seismic energy is released (Endo and Murray, 1991)--were roughly constant. Beginning on June 1, however, RSAM values began to increase and exhibited greater variation. Tremor episodes became more frequent and more intense, and on June 6 the locus of earthquake activity shifted away from the region to the northwest. At about 0700, seismicity increased rapidly beneath the fumaroles just north of the summit; many of these earthquakes were quite shallow. This period of intense seismicity lasted until 2300 on June 7, when it abruptly diminished (Harlow and others, this volume).
A tiltmeter located on the east flank of Mount Pinatubo detected accelerating inflation on June 6, at about the same time as intense seismicity began beneath the fumaroles (Ewert and others, this volume). This inflationary event ended abruptly at 1700 on June 7, when a sudden gas and ash emission produced a plume whose top reached about 4 km over the volcano.
Figure 3. Steam and ash emission in the uppermost Maraunot valley on May 30, 1991. (View is approximately to the southeast.) Arrow shows approximate site where a lava dome appeared on June 7. (Unless noted otherwise, photographs by R.P. Hoblitt.)
The emergence of a lava dome provided the first unequivocal evidence that Pinatubo's activity was driven by rising magma. The inception of dome growth was not witnessed directly. The incipient dome was first sighted at about 0800 on June 8; it was not present when Pinatubo was inspected the previous morning, a time when viewing conditions were good. We conclude that dome growth began late on June 7, probably between 1700, when inflationary tilt ceased, and 2300, when intense seismicity beneath the fumaroles rapidly declined.
Details concerning dome growth are meager because heavy ash emission hindered observation during helicopter inspections. When first sighted (fig. 4), the dome clung to the east wall of the upper Maraunot River canyon. The emergence of the dome at this location (fig. 3) was unexpected because the closest previous activity was a weak fumarole about 100 m to the east. Several vigorous fumaroles were in the general area, the most vigorous of which was about 300 m to the south, in the bed of the uppermost Maraunot River (fig. 3).
Ash emissions, most of which appeared to originate at the east margin of the dome, waxed and waned. In general, however, periods of elevated ash emission were becoming more frequent, and the emission rate during these periods appeared to be increasing. On June 9, ash emission was at times sufficiently heavy to produce ash curtains (fig. 5). During the morning and afternoon of the same day, observers on the west side of Pinatubo (Sabit and others, this volume) reported "pyroclastic flows" moving down the west and northwest flanks of Pinatubo. Subsequent helicopter inspections showed these areas to be dusted with ash but devoid of the destroyed vegetation or deposits typical of pyroclastic flows or surges. Apparently the ash content of the plume occasionally reached concentrations sufficient to produce density currents whose densities marginally exceeded that of the ambient air.
The apparent increase in the rate of ash emission was confirmed by the C-band weather radar at Clark. On June 10, a visually impressive ash cloud was not detected by the radar, but at 1040 on June 11, the ash concentration exceeded the detection threshold.
SO2 emissions, which began to increase just before the appearance of the dome, continued to increase (Daag, Tubianosa, and others, this volume). The flux rose from about 1,500 t/d on June 8 to >10,000 t/d on June 10, the last SO2 measurement before the climactic eruption.
Dome growth continued at least until 0700 on June 11--the last inspection during Phase III (fig. 6). Sketch maps of the dome's perimeter on June 8, 9, and 11 (fig. 7A-C) document the growth. Our estimates of the mean dome diameter and height are 200 m and 40+-10 m, respectively, on June 11, for a volume of about 1-1.5 x 106 m3. We are uncertain whether extrusion was continuous or episodic. However, episodic seismicity during the period of dome growth (Harlow and others, this volume; Power and others, this volume) suggests that dome growth may have also been episodic.
Once magma reached the surface, seismicity diminished to a low level until the evening of June 8. Seismicity through the balance of Phase III--that is, until June 12--was characterized by swarms of volcano-tectonic earthquakes and gradually increasing tremor, which became continuous on June 10 (Harlow and others, this volume; Power and others, this volume). Overall, the seismic energy release was escalating.
The dome was not sampled directly, because of safety considerations. Samples obtained after the climactic eruption show the dome lava to be andesite, produced by mixing basalt and dacite magmas (Pallister and others, this volume).
Figure 4. The dome as it was first sighted on June 8, 1991, on the east side of the upper Maraunot valley. (View is approximately to the south.) (Photograph by Valentino M. Gempis, U.S. Air Force.)
Figure 5. Curtains of falling ash on June 9, 1991. (View is approximately to the west.)
Figure 6. The dome (indicated by arrow) on June 11, 1991, the last view before the June 12 eruption. The dome's west margin has reached the bottom of Maraunot valley. (View is approximately to the west.)
Figure 7. Sketches of the dome's perimeter. A, June 8. B, June 9. C, June 11. D, June 14. Vertical eruptions of June 12 and 13, from a vent near the dome's southeast margin, removed part of the dome and produced the crescent shape seen on June 14. Contour interval is 20 m.
A small explosion at 0341 on June 12 initiated Phase IV. The event was detected seismically at Clark (Harlow and others, this volume) and visually from Poonbato (fig. 1), on the west side of Pinatubo (Sabit and others, this volume). At about 0400, Clark radar detected a cloud over Pinatubo at an altitude of about 11 to 12 km; however, the presence of large thunderheads in the vicinity of Pinatubo at this time makes interpretation of the radar observation ambiguous. At Clark, dawn (~0500) revealed a white plume rising from the volcano. Radar reflections at 0540 were described as the strongest yet seen; they placed the plume top at an altitude of about 3.5 to 4.5 km. The vigor of the plume declined, and by 0630 it was apparent that the plume originated at two closely spaced vents. During a helicopter inspection about 30 min later, one source was seen to be the east margin of the dome, while the other was at the site of the most vigorous predome fumarole, in the bed of the uppermost Maraunot River (fig. 8). The east and north flanks of the volcano were heavily dusted with ash, at least some of which must have been transported as a weak density current, because the boundary between ash-covered and unaffected terrain was sharp to the east (fig. 9). No unequivocal pyroclastic-flow deposits were recognized in the drainage that originated on the east flank (Sacobia); however, the upper parts of the O'Donnell and Maraunot drainages contained minor pyroclastic-flow deposits that gave way downstream to hot lahar deposits (fig. 10). By 0730, the vigor of Pinatubo's plume had declined further, and the ash content appeared to be low, but ash emission still precluded an inspection of the dome.
The first large vertical eruption began at 0851 on June 12. As seen from Clark, the developing plume was rather broad, with two main protuberances, perhaps a reflection of the two vents noted just before the eruption. Initially, the northern protuberance was dominant, but it soon became subordinate to the southern one (fig. 11). The plume top eventually exceeded the height range of the Cubi Point radar (19 km); Clark radar operators were evacuated from their station when the plume top was at 19 km and still rising. As the eruption continued, a circular collar expanded around the plume at the tropopause (fig. 12; Oswalt and others, this volume).
Our reconnaissance helicopter reached a good vantage point to the north of Pinatubo at about 0945. The course of the upper Maraunot River was marked by diffuse ash clouds that were punctuated in a few places by roiling clouds; apparently, secondary explosions in the river were feeding ash into the adjacent diffuse clouds. At about 0950, a new density current began to move to the north, down the O'Donnell River (fig. 13), and probably down the Maraunot as well. This current dissipated by 1005, after moving down the O'Donnell about 4 km from the vent, and an unknown distance down the Maraunot. We left the area about 1010 to refuel and returned at about 1040. Activity at the vent had now declined to a low level. Secondary explosions in the Maraunot (fig. 14) suggested that pyroclastic flows reached at least 4 km from the vent in that drainage; the explosions and associated ash clouds prevented direct observation of the flow deposits. Explosions were not occurring in the O'Donnell. Pyroclastic-flow deposits were absent there, as was any other evidence of their passage. The upper few kilometers of that drainage were, however, heavily dusted with ash, the only apparent product of the density current we witnessed during the previous trip. The density current must have been a slack surge. Upon inspecting other drainages it became apparent that only the Maraunot had received significant pyroclastic-flow deposits during the eruption. The axis of the tephra deposit crossed the southwest flank of the volcano.
Figure 8. Emissions from Mount Pinatubo at about 0700 on June 12. Plume on the left is from the strong, persistent vent to the south of the dome (compare fig. 3); diffuse plume on the right is from the dome. (View is approximately to the south.)
Figure 9. The east flank of Mount Pinatubo, in the upper drainage of the Sacobia River, showing a margin of the area affected by a weak, ash-laden density current, probably generated by the 0341 explosion on June 12. Photograph taken shortly before 0700.
Figure 10. Hot lahar deposit in the bed of the Maraunot River at about 0700 on June 12. Note steam rising from hot boulders.
Figure 11. Plume from the eruption that began at 0851 on June 12. (View is to the west from Clark Air Base at 0900.) The plume protuberance to the right (north) was initially dominant but was overtaken by the protuberance to the left.
Figure 12. Plume from the eruption that began at 0851 on June 12. (View is from the north at 0945.) Note the circular collar (indicated by arrow) that developed around the plume where it penetrated the tropopause.
Figure 13. Pyroclastic density current moving to the north of Mount Pinatubo, down the O'Donnell River (center) at about 0950 on June 12. Current probably also moved down the Maraunot River (ash-covered, right). (View is from the north.)
Figure 14. Secondary explosions in the Maraunot River at 1140 on June 12. (View is to the southwest.)
When the 0851 eruption began, high-amplitude eruption signals abruptly ended a period of relative seismic quiescence that began about 3.5 h earlier. The seismically intense part of the eruption lasted 36 min; however, as noted above, visible activity at the vent persisted until at least 1000. In general, the level of seismicity after the eruption was substantially greater than before the eruption (fig. 2A), largely due to the appearance of energetic long-period events (Harlow and others; Power and others, this volume). A swarm of long-period earthquakes began about 1725 and lasted about 50 min.
The second vertical eruption began at 2252, about 14 h after the first. On the basis of seismicity, it lasted about 14 min. Like the first eruption, it began abruptly--ending 4.5 h of relative seismic quiescence (fig. 2A). Clark radar indicated the eruption-column height exceeded 24 km. Tephra was again dispersed to the southwest, this time in the midst of scattered thunderstorms. Direct observations were scanty because of darkness. Abundant lightning was witnessed in the vicinity of Pinatubo, from Clark and from Camp O'Donnell, about 30 km to the north, but at least some this was probably due to the thunderstorms rather than to the eruption. Another swarm of increasingly energetic long-period events began about 0600 the following morning (June 13) and lasted roughly 2 h. We were unable to inspect the volcano before the next eruption began.
A third vertical eruption began in clear weather at 0841 on June 13 (fig. 15) about 10 h after its predecessor. It lasted only about 5 min, on the basis of seismicity. The onset of the eruption was again abrupt, but the period of relative quiescence that preceded it was only about 15 min, much less than those of the first two eruptions (fig. 2A). Clark radar indicated the column height again exceeded 24 km, although the radar reflections were weaker than those of the second eruption. As before, tephra was dispersed to the southwest.
An inspection by helicopter at about 1000 was hindered by a thick haze that hung over Pinatubo. However, we did observe a small pyroclastic-flow deposit (fig. 16) in a ravine in the uppermost Sacobia drainage. It is uncertain whether this flow was the product of the second or third eruption. In any case, it was apparently the first to be emplaced on the east flank of the volcano.
About 3 h of relative seismic quiescence followed the third eruption. Then, at about noon on June 13, a swarm of small long-period events began, occasionally punctuated with larger events. The intensity of the swarm increased gradually until about 0400 on June 14, when the intensity began to increase rapidly. Dawn brought excellent viewing conditions and revealed that, remarkably, steam and ash emissions from Pinatubo were essentially nil (fig. 17). Convinced that an eruption was imminent, we postponed aerial inspection. But the intense seismicity persisted hour after hour, and viewing conditions from Clark slowly deteriorated. Concerned that approaching Typhoon Yunya might prevent observations for an extended period of time, we began to cautiously inspect the volcano at about 1215.
The west margin of the dome now nearly spanned the narrow Maraunot River canyon (fig. 18). The vent--dimly visible through ash--had taken a semicircular bite out of the southeast margin of the dome (fig. 7D). The vent diameter was approximately 200 m. New pyroclastic-flow deposits extended down the Maraunot River valley as much as 4.5 km from the vent (figs. 18, 19); the O'Donnell may have received small pyroclastic flows, but, if so, they extended less than 1.5 km north of the vent. The absence of substantial pyroclastic-flow deposits in the O'Donnell drainage is somewhat surprising, because only a low interfluve separated the uppermost O'Donnell drainage from the vent. Clouds prevented us from inspecting the south and west flanks; however, the dearth of pyroclastic-flow deposits in the O'Donnell suggests that the south and west drainages were also little affected by pyroclastic flows.
The swarm of long-period events culminated in the fourth vertical eruption at 1309 (fig. 20). This eruption, in contrast to the previous three, was not preceded by a period of relative seismic quiescence. The eruption lasted only 3 min, then seismicity dropped to low-amplitude tremor. Radar at Clark indicated a column height of 21 km; tephra was again carried to the southwest. The relatively low plume height supports our subjective impression: the 1309 eruption was the least vigorous of the four. This was unexpected, as we mistakenly assumed that the intense precursory seismicity would culminate in a large eruption.
Figure 15. Plume from the eruption that began at 0841 on June 13. (View is to the west from Dau on Clark Air Base at 0848.)
Figure 16. The margin of an area affected by a pyroclastic density current produced by the 2252 eruption of June 12 or the 0841 eruption of June 13; photograph taken at 1006 on June 13. (View is to the west at the east flank of Mount Pinatubo, in the uppermost Sacobia drainage.) Smoke rises from vegetation incorporated into a hot deposit in the bottom of the ravine.
Figure 17. Dearth of steam and ash emission from Mount Pinatubo on the morning of June 14, during a period of intense seismicity. (View is from Clark Air Base to the west at 0722.)
Figure 18. The dome (center) at 1220 on June 14, as seen from the northwest, up the Maraunot River canyon. New pyroclastic-density-current deposits (lower center, light-colored) sit in the canyon bottom and mantle the adjacent terrain.
Figure 19. Newly emplaced pyroclastic-flow deposits partially filling the canyon of the Maraunot River about 5 km from the old summit (site 24, fig. 24). (View is to the southeast at 1221 on June 14.)
Figure 20. Plume from the eruption that began at 1309 on June 14; (View is to the west from the southwest end of the airfield at Clark Air Base at 1324.)
The pace of eruptive activity increased after the 1309 June 14 eruption, just as viewing conditions deteriorated with the approach of Typhoon Yunya. Opportunities for direct observation became progressively less frequent, particularly from Clark, so we became more dependent on instrumental data. We emphasize here the scanty direct observations, and refer the reader to table 1 for some instrumental observations not described in the text.
Following the 1309 eruption, we flew to a site 15.5 km north of Pinatubo in an attempt to repair seismic-telemetry gear and arrived at about 1410. Viewing conditions were poor; however, it was clear that the volcano was feeding copious quantities of ash into a broad plume, and ash clouds were rising from multiple sources in the vicinity of the Maraunot and O'Donnell drainages. Other drainages were obscured. Apparently, small explosions or low fountaining generated density currents from which the ash clouds were rising. Radar at Clark detected the ash, which reached a peak altitude of about 15 km at 1411. Winds from Typhoon Yunya carried the ash eastward, along with a thunderstorm that grew amidst the ash. This was to be the first ash to reach Clark, where it fell mixed with rain during the afternoon. Unbeknownst to us, this inspection was to be our last before the climactic eruption. We left the site at about 1435.
This event could be considered either a vestige of the 1309 eruption, a small discrete eruption, or simply a precursor to the larger eruption that occurred a short time later. The start time is uncertain, though it may have been at about 1400, when a modest intensification in the low-amplitude tremor occurred. Seismicity provides no basis upon which to estimate a duration.
A swarm of long-period events began at 1438 and quickly intensified. At about 1516 a large pyroclastic density current swept out radially from Pinatubo and reached a distance of about 15 km to the west-northwest. Observers to the west had to flee. This behavior differs markedly from that of the Phase IV eruptions, which produced high, relatively narrow eruption columns along with a few channelized density currents of limited extent. The event was not visible from Clark because of the ashy thunderstorm that developed earlier, and, probably for the same reason, Clark radar did not recognize the event as an eruption. Cubi Point radar, however, reported a tephra cloud between 1530 and 1600, moving southwestward, with a top at 18 km. Hot lahars that moved down the Sacobia River past Clark that evening were almost certainly triggered by pyroclastic debris deposited into the Sacobia drainage by this eruption.
The 1516 eruption produced a sizable deflection on the Clark barograph (fig. 2C). The barograph deflections are peculiar to the Phase V eruptions; the vertical eruptions of Phase IV did not produce them. At least 13 barograph deflections were recorded during Phase V, and at least five of the eruptions that produced them were witnessed.
The second large Phase V eruption began at 1853 and produced a plume whose height exceeded 24 km. The eruption was not observed directly. The next two eruptions, which began at 2320 on June 14 and 0115 on June 15, were observed from Clark with the aid of an infrared imaging device. The infrared images show ground-hugging density currents moving north, south, and east; density currents presumably also moved westward.
At 0555 on June 15, just after dawn, another eruption produced a radial pyroclastic density current. This eruption was clearly visible from Clark (fig. 21). The density current was driven by a pyroclastic fountain whose projections were briefly visible at the start of the eruption but were soon concealed by ash clouds rising convectively from the density current. This fountain resembled those observed during the postclimactic eruptions of Mount St. Helens in 1980 (Hoblitt, 1986). Ash clouds from throughout the affected area merged into a great cloud that rose over the volcano; briefly, the periphery of this cloud was laced with orange-red lightning. As the cloud rose, surface winds swept up ash from the newly emplaced deposits and carried it back toward the volcano. Radar indicated a maximum plume height of 12 km. A diffuse, light-gray, ground-hugging cloud then formed in the vicinity of Pinatubo and seemingly moved out radially (fig. 22); it certainly moved eastward, because it arrived at Clark about 40 min after the eruption began. Ash mixed with rain began to fall as the cloud arrived. Pinatubo was not seen again from Clark until after the climactic eruption.
For a time, Pinatubo remained visible from the north and west. Another density current was witnessed at about 0810 on June 15 from both Camp O'Donnell and Poonbato (fig. 1); yet another began at 1027 and was witnessed from Poonbato. Poonbato observers regarded the 1027 density current as the largest that they witnessed--marginally larger than the 1516 event of June 14. The last density current to be witnessed began at 1117 on June 15. It was dimly visible from Poonbato and Camp O'Donnell. Subsequent density currents were only detected instrumentally and confirmed at Clark by periods of nearly complete darkness beginning about 35 to 40 min after each eruption; darkness was due to the arrival of wind-drifted ash. Subsequent stratigraphic studies, described below, demonstrate that the Phase V density currents were pyroclastic surges. Accordingly, we refer to the Phase V events as "surge-producing" eruptions.
Figure 21. Photographs of the pyroclastic surge that began at 0555 on June 15, 1991. A, 0558. B, 0601. (View is to the west from Dau complex on Clark Air Base.)
Figure 22. A diffuse, light-gray, ground-hugging cloud approaching Clark at 0622 on June 15. The cloud formed in the vicinity of Mount Pinatubo after the 0555 pyroclastic surge and apparently moved out radially. (View is to the west from Dau on Clark Air Base.)
Some reconnaissance work was conducted in late June and early July of 1991. The new deposits were nearly undissected, so outcrops were available only in the thinner upland or distal deposits. Because of this and safety considerations, most early work was restricted to brief examinations of thin distal deposits.
The majority of fieldwork was conducted in February and March of 1992. By this time, deposits had been subject to one rainy season (July-October) since their emplacement. Access to the proximal deposits was difficult because all access roads were damaged or destroyed by lahars generated by erosion of the fresh deposits during the rainy season. The termini of pyroclastic-flow deposits could be reached by road in the Sacobia, Pasig, and O'Donnell drainages, but further travel on foot was impeded by a remarkably well-developed dendritic erosion pattern. The new deposits had been sculpted into a badlands topography of steep-sided rills and canyons that provided excellent exposures but was exceedingly difficult to traverse on foot. The majority of fieldwork on proximal deposits was conducted during 7 days when helicopter support was available. In an attempt to locate sections containing deposits from as many of the eruptions as possible, we focused on the region to the southwest of the volcano, where tephras from the June 12-14 eruptions were deposited. Areas of interest were identified on aerial photographs, but the actual study sites were selected from the air, as the helicopter circled the area. Once on the ground, because of restricted mobility and limited time, we could explore only a rather small area. Our results are, therefore, only preliminary.
All proximal sites that we examined lie within the area affected by pyroclastic flows from the climactic eruption. These flows extensively scoured the surfaces over which they passed: consequently, the Phase IV fall deposits and the overlying Phase V surge deposits were found only as isolated erosional remnants beneath the deposits of the climactic eruption (Phase VI). Such remnants were preserved in topographically protected areas, such as depressions or in the lee of obstacles.
Andesite, dacite, and accessory component proportions were determined from >4-, >8-, or >16-mm clast fractions by hand sorting them with the aid of a binocular microscope. Andesite and dacite were identified by the presence and absence of clinopyroxene.
Clasts selected for density analysis were washed in water, dried overnight at 110°C, sprayed with an aerosol containing a trace amount of silicon oil, dried overnight at 110°C, weighed in air, and weighed in water. Bulk volume was calculated from the weight difference by using Archimedes' principle.
Because deposits were examined and sampled at numerous localities, it is convenient to refer to a composite stratigraphic section (fig. 23). This section also provides the rationale for the deposit sequences used in plots of clast density and component proportions. We include Phase VI fall and flow deposits; these are discussed in detail by W.E. Scott and others (this volume), and Paladio-Melosantos and others (this volume).
Figure 23. Diagrammatic stratigraphic sections illustrating relations between Phase IV, V and VI units sampled for this study. Site 24: section in Maraunot River canyon (see figs. 24, 29). Southwest proximal composite: Phase IV fall deposits from site 22 (see figs. 24, 25, 26); Phase V surge deposits generalized from sites 11, 17, 20, 21, 22, 23 (see figs. 24, 30); Phase VI fall deposits from site 17 (see W.E. Scott and others, this volume, fig. 19), Phase VI pyroclastic-flow deposits generalized for proximal drainages. Distal composite: generalized distal stratigraphy (see, for example, fig. 31), Phase IV fall deposits lie only to the southwest. Dashed line shows phase boundaries, dotted line shows intraphase correlations.
The best example of the Phase IV fall deposits was found at site 22 (figs. 23-26), where their total thickness reached 14 cm. Assignment of eruption times to the various beds was aided by the fact that the second eruption of June 12 (beginning at 2251) occurred during a rainstorm. Cohesion of moist particles (rainflushing) produced a fall deposit that is finer grained and more poorly sorted than those above and below it (table 2). The section was sampled for granulometric, component, and density analysis as indicated on figure 26; note that the thickness of the beds decreases upwards and that the 1309 June 14 eruption is represented only by a thin ash bed, which was not sampled. The sampled fall deposits are dominantly andesite, although the proportion of dacite increases upward (figs. 27 and 28).
Pyroclastic-flow deposits emplaced during the Phase IV vertical eruptions were examined at one site in 1992 (fig. 24, site 24; fig. 23), in the canyon of the Maraunot River. This site received density currents--because the vent was located in the uppermost Maraunot drainage--but stood north of the area that received fall deposits. The lower part of the Phase IV section consists of a thin (10-30 cm) pyroclastic-flow deposit sandwiched between thin ash beds: two ash beds below, three above (fig. 29A). The pyroclastic-flow deposit was almost certainly emplaced sometime during the first (0851) June 12 eruption because, after the eruption, we observed secondary explosions in the Maraunot drainage that extended to the vicinity of site 24 (fig. 14). Component and density data also support correlation of this deposit with the 0851 event (figs. 27 and 28). We interpret the thin ash beds as the products of slack surges of the sort reported by west-side observers on June 9 (Sabit and others, this volume) and the sort we observed during the latter part of the 0851 eruption (fig. 13).
The upper part of the site 24 section is dominated by two thick (3-4 m) pyroclastic-flow deposits (fig. 29B). The lower one is rich in dense, prismatically jointed blocks of andesite; these are interpreted as fragments of the dome that began to grow on June 7 (Pallister and others, this volume). This must be the pyroclastic-flow deposit photographed in the Maraunot on June 14 (fig. 19); it must have been emplaced during the second (2252) eruption of June 12 or during the June 13 eruption (0841).
The uppermost pyroclastic-flow deposit at site 24 consists of normally graded lithic debris in a pumiceous matrix. Similar deposits are found in all of Pinatubo's major drainages, always at the top of the section produced during the climactic eruption (Phase VI). They were probably produced during caldera collapse (W.E. Scott and others, this volume). Numerous preclimactic events were not recorded at this proximal site (fig. 23), probably because of the erosive nature of the Phase VI pyroclastic flows.
Figure 24. Locations of sites where Phase IV and V deposits were examined. Dashed line shows approximate margin of area affected by pyroclastic flows; dark-gray shading shows thick, ponded pyroclastic-flow deposits (see W.E. Scott and others, this volume). Heavy line outlines caldera; white triangle, old summit; white star, lava dome of June 7-15. Site numbers are the same as those in W.E Scott and others (this volume).
Figure 25. Deposits of June 12-16 at site 22. The Phase IV fall deposits lie directly on soil. The fall beds are overlain successively by surge deposits, fines-poor basal facies of the climactic pyroclastic-flow deposit, and the climactic pyroclastic-flow deposit. Length of ruler is 15 cm.
Figure 26. The various preclimactic fall deposits at site 22 (figs. 23, 24). Vertical lines show how units were sampled for component and density analysis; the deposit from the June 14 eruption was not sampled.
Figure 27. Normalized frequency of juvenile andesite and dacite clasts for two composite stratigraphic sections from the June 1991 eruptions. A, Composite section including component data from surge beds at site 23. B, Composite section including component data from surge beds at site 11. The proportion of juvenile basalt, a minor component associated with andesite, is not shown.
Figure 28. Density distributions of lapilli from various stratigraphic units produced by the June 1991 eruptions. Graphs arranged in the stratigraphic sequence used in figure 27A. A, The lower of two preclimactic pyroclastic-flow (PF) deposits at site 24. B-D, Sample fractions A, B, and C (site 22, fig. 26) of the June 12, 0851 fall deposit. E, Fall deposit from the June 12, 2252 eruption (site 22, fig. 26). F, Fall deposit from the June 13, 0841 eruption (site 22, fig. 26). G, The upper of two preclimactic pyroclastic-flow deposits at site 24. H-L, Five sequential surge beds from site 23. M-P, Four sequential sample fractions from the June 15 climactic fall deposit at site 17.
A-B, C-D, E-F, G-H, I-J, K-L, M-N, O-P
Figure 29. June 1991 strata at site 24; see figure 23 for a diagram of stratigraphic relations. A, Phase IV strata at the base of the June 1991 section: a thin (10-30 cm) pyroclastic-flow deposit (behind shovel handle) sandwiched between five thin ash beds (three above, two below). These units are underlain by alluvium and overlain by the two pyroclastic-flow deposits shown in B. B, Two (3- to 4-m thick) pyroclastic-flow deposits that compose the upper part of the June 1991 section. Lower flow (Phase IV) is rich in dense, prismatically jointed blocks of andesite interpreted as fragments of the lava dome that began to grow on June 7; the same flow deposit, undissected, is shown in figure 19. Upper flow consists of normally graded lithic debris in a pumiceous matrix, probably produced during caldera collapse in Phase VI. Minor resistant bed that caps the right side of the section may be a posteruption debris-flow deposit.
Table 2. Grain-size parameters for deposits produced by the preclimactic eruptions.
We found remnants of the Phase V deposits at six (sites 11, 17, 20, 21, 22, 23; fig. 24) of the eight proximal sites that we visited. Phase V deposits (figs. 23, 30A,B) are composed of multiple surge beds, each of which is readily divisible into a light-gray relatively coarse-grained lower layer and a gray, brown, or pink-brown relatively fine-grained upper layer. The fine-grained upper layers resemble the distal Phase V beds that are described below. Individual beds (coarse layer + fine layer) attain a thickness of as much as a few decimeters; the total thickness of Phase V deposits ranged up to about 1 m at our most proximal site (fig. 24, site 11; fig. 30B). In the thicker beds, the coarse-grained layer consists of a massive, matrix-supported facies that may locally grade upward into a finer grained stratified facies or may grade downward to a friable, fines-deficient, grain-supported facies. All three of the coarse-grained facies were present in a few well-developed examples. Thicker beds exhibit low-angle crossbedding, dune structures, pinch-and-swell structures, or plane-parallel bedding. The coarse-grained layer is in sharp contact with the overlying fine-grained layer, which typically contains accretionary lapilli. Thinner beds consist only of the fine-grained layer or of the stratified facies of the coarse layer and the fine-grained layer. The lower few beds typically contain organic debris, some of which is carbonized. In most sections we examined, the lowermost surge bed is thicker and coarser than those that overlie it. In the most complete sections, the thickness and coarseness of the sequence tend to decrease upwards.
In the exposures we examined, the top of the Phase V sequence almost always consists of one of the fine-grained layers; this is overlain by climactic (Phase VI) pyroclastic-flow deposits. As the contact between the two phases is traced laterally, it typically jumps to a stratigraphically lower or higher fine-grained layer. This jump implies that surge beds were locally scoured away by the climactic pyroclastic flows. Because of this nonconformable contact, we are uncertain whether any of the proximal surge sections we examined are complete. The maximum number of depositional events estimated for any proximal section is 11; 13 surge-producing eruptions were inferred from the barograph record (fig. 2C).
Figure 30. Proximal Phase V surge beds overlain by Phase VI climactic pyroclastic-flow deposits at (A) site 23 and (B) site 11. Subdivisions on the left side of the scale are centimeters.
Distal surge deposits typically consist of multiple beds of gray, pinkish-brown, and brown silt-sized ash (fig. 24, site 5; fig. 31). Most bedding is planar in the distal sections, though crossbedding, normal grading, and cut-and-fill structures are locally present in individual beds as much as 14.5 km from the old summit. Individual bed thicknesses range up to a few centimeters. Some beds contain accretionary lapilli, and, locally, some of the lower beds contain uncharred plant debris, dominantly leaf fragments (fig. 31B). Void spaces are locally common in some beds; some of these have irregular margins and constitute the interstices between accretionary lapilli (fig. 31C). Other voids have rounded margins and occur within otherwise structureless ash (fig. 31C). Some contacts are sharp, while others are transitional; thus, it is difficult to establish the exact number of depositional events represented in a given section. Some beds are rather uniform throughout and are delineated by sharp contacts; each of these probably represents a single depositional event. Other beds are graded, either normally or inversely, with respect to grain size or the abundance of accretionary lapilli. At least some of these graded beds are probably the products of more than one depositional event. Even with the difficulty in counting beds, the number of surge beds clearly varies from outcrop to outcrop but tends to increase with increasing proximity to the volcano. The maximum number of depositional events estimated in any distal section is 10.
To the southwest of Pinatubo, the surge deposits overlie the fall deposits from the June 12-14 vertical eruptions; in other sectors they lie directly on the preexisting surface. The distal surge deposits are everywhere overlain by fall deposits from the June 15 climactic eruption.
Figure 31. Distal Phase V surge deposits. A, Surge sequence (behind ruler; length, 15 cm) at site 5, overlain by fall and ash-cloud deposits from the climactic eruption. B, Lower part of the surge sequence at site 5. Note leaf and other organic material in the unit just above the soil. C, Closeup view of the section shown in B. Note accretionary lapilli in the upper unit and spherical voids in the lower units.
Spherical voids of the sort present in some of the fine-grained surge beds (fig. 31B,C) are commonly interpreted as evidence of deposition of hot, wet ash. This interpretation is warranted if there is also evidence of cohesion during deposition--for example, if the deposit is plastered onto vertical surfaces. In our experience, spherical voids can also form by postdepositional wetting of ash deposited at low temperatures in a dry, expanded condition. Vertical surfaces, such as tree trunks, were numerous at the distal sites that we visited, but they were not plastered with ash. Thus, voids like those in figure 31C probably formed when dry, powdery ash deposits were moistened after deposition. We are uncertain how the voids form but speculate that the passage of a saturation front through the loose, dry ash realigns particles into a close-packed configuration. The spherical voids reflect the change in interstitial volume as loose, dry ash is converted to compact, wet ash.
From their stratigraphic position, the coarse-grained surge deposits examined in the proximal area must be the products of pyroclastic fountains of the sort we witnessed during the 0555 eruption of June 15. The area swept by these energetic fountain-fed surges is only known approximately. Direct observations from the ground were limited because events occurred at night or in poor weather. For these same reasons, and because winds of Typhoon Yunya deposited ash on Clark Air Base, observations from aircraft were not possible. Were it not for the subsequent pyroclastic flows of the climactic eruption, the margin of devastated vegetation could be used to delineate the area. But in many sectors the pyroclastic flows were as extensive or more extensive than the surges. Photographs and eyewitness accounts provide a few constraints. Photographs taken by the first author at Dau (Clark), when compared with a videotape taken by Maj. Keith McGuire from Camp O'Donnell, indicate that the 0555 surge of June 15 reached a point about 10 km north of Pinatubo's former summit, in the vicinity of the headwaters of the Bucao River. This surge moved about 7 to 8 km to the east, down the Sacobia River valley to the vicinity of Mount Dorst. Its extent to the south was less, probably no more than 2 to 3 km from the old summit. A videotape of this event was taken from the northwest by one of us (Dindo Javier). Although the travel distance cannot be determined from the tape because of the dearth of topographic markers, eye witnesses estimate that the surge reached within 1 to 2 km of Burgos, about 14 km from the old summit.
Less information is available for the other surges. Witnesses reported that the (first) surge at 1516 on June 14 reached about the same distance to the northwest as the 0555 event. As evidenced by blasted trees, the presence of surge deposits, and the absence of pyroclastic-flow deposits at a site about 1 km north of Mount Cuadrado, the surges collectively reached a maximum distance of about 10 km to the south of the old summit.
The fine-grained facies of the surge deposits, which is found in proximal as well as distal areas, is far more extensive than the coarse-grained facies. Its extent may be appreciated by inspecting an isopach map of tephra layer B (Paladio-Melosantos and others, this volume; fig. 6). Layer B is defined as tephra deposited after the first June 12 eruption and before the climactic eruption of June 15; thus it contains contributions from both Phase IV and V. However, the contribution from Phase IV is only significant to the southwest of Pinatubo.
In this section we discuss topics relating to the origin of the Phase V surge deposits and to patterns evident in the preclimactic eruptions.
The Pinatubo surge deposits are similar in many respects to deposits produced by the directed blast at Mount St. Helens, Wash., in 1980 (Hoblitt and others, 1981; Moore and Sisson, 1981; Fisher, 1990) and by the climactic eruption at Mount Lamington, Papua New Guinea, in 1951 (Taylor, 1958). All three deposits are the products of pyroclastic density currents, have low aspect ratios, generally become thinner and finer grained with distance from source, and, at least locally, exhibit features associated with surge deposits. Furthermore, all exhibit similar granulometries (fig. 32A).
There are also differences. The Pinatubo deposits were produced by numerous events separated by tens of minutes to hours--the Mount St. Helens and Mount Lamington examples by single events or, in the case of Mount St. Helens, perhaps by two events separated by tens of seconds to a few minutes. The Mount St. Helens event was laterally directed, the consequence of sector collapse exposing a cryptodome. The Mount Lamington and Pinatubo examples involved vertically directed eruption fountains that collapsed to form radially directed surges. The Mount St. Helens event devastated an azimuthally restricted area of about 600 km2, Mount Lamington devastated a radially distributed area of about 230 km2, and the Pinatubo events collectively devastated a radially distributed area of about 300 km2. The source mechanism for the Pinatubo surges resembles that of Mount Lamington more than that of Mount St. Helens.
Figure 32. Median grain size (Mdf) versus sorting (sf) for Phase V surge deposits. A, Plot comparing Pinatubo data to that of the 1980 Mount St. Helens and 1951 Mount Lamington (Papua New Guinea) surge deposits. B, Proximal data, site 11. Lines connect upper ash layer (filled squares) and lower coarse-grained facies (open squares) from individual surge beds. C, Proximal data, site 23. D, Data from both distal and proximal surge deposits.
Because of the similarity of their grain-size distributions (fig. 32B-D; table 2), the distal Phase V deposits appear to be equivalent to the upper fine-grained facies of the proximal Phase V beds. We know from direct observation that the distal beds were produced dominantly by subvertical fallout from slack surges. Because the distal beds and the fine-grained facies of the proximal beds are granulometrically indistinguishable, and because accretionary lapilli are common in both, we suggest that both were emplaced by the same depositional process. That is, both were emplaced by slack surges whose progenitors were the energetic fountain-fed surges that emplaced the lower coarse-grained facies of the proximal Phase V beds.
The Phase V surges caused atmospheric-pressure disturbances that were recorded on barographs at Clark Air Base (fig. 2C) and Cubi Point (Oswalt and others, this volume). Interestingly, the four vertical eruptions of Phase IV did not produce deflections. Most of the surges produced an impulsive compression, followed by a more sustained rarefaction, followed by a second even more sustained compression (fig. 2C). This same behavior was documented for the 1980 lateral blast at Mount St. Helens (Reed, 1980; Banister, 1984). The similarity of the Mount St. Helens and Pinatubo barographs rather strongly suggests that the same mechanisms were operating at both volcanoes. Banister (1984) attributed the initial compression at Mount St. Helens to the lateral explosion and attributed the following sustained rarefaction and compression to the subsequent plinian eruption.
In each of the Pinatubo examples, it seems likely that the initial compression was due to the formation of a fountain-fed pyroclastic surge. Two processes probably contributed to the transient atmospheric compression: (1) the sudden ejection and decompression of eruption products and (2) incorporation and expansion of air as the surge engulfed terrain around the vent. Because the barographs did not record pressure deflections during the Phase IV vertical eruptions, we conclude that the second process--incorporation and expansion of surface air--was more important than ejection and decompression. Expansion of incorporated air and deposition of suspended pyroclasts progressively lowered the density of the surge clouds until they became buoyant and lifted off the ground (Sparks and others, 1986). This great volume of ash-laden gas then rose diapirically. As the diapir rose, surrounding air rushed in to replace it; this probably caused the barometric rarefaction. It seems likely that the second barometric compression was due to the formation of diffuse density currents (slack surges) of the sort that reached Clark about 30 to 40 min after the start of surge-producing eruptions (fig. 22). We can only speculate on the process that produced the slack surges. Perhaps the buoyant mass of ash-laden gas behaves somewhat like a thunderstorm (Barry and Chorley, 1978); that is, it cools adiabatically until condensation occurs. Precipitation in the form of accretionary lapilli would produce a downdraft of relatively cool, ash-laden air that would flow outward as a density current (the slack surge), after it reached the ground. This would account for the abundance of accretionary lapilli in the slack surge deposits.
Whatever the process responsible for the slack surge, it apparently also operated at Mount St. Helens following the lateral blast of May 18, 1980. This is evident from photographs and eyewitness accounts presented by Foxworthy and Hill (1982, p. 56-57). We suggest that this slack surge, rather than the development of a vertical eruption column, was responsible for the second barometric compression observed in the Mount St. Helens example.
We envision three stages to each Phase V event: (1) development of an energetic fountain-fed surge that swept out radially from the vicinity of the vent until its density dropped below that of ambient air; (2) buoyant rise of a great diapir of ash and hot gas, coupled with a "back draft" as air flowed inward to replace that which had risen; and (3) development of a slack surge beneath the diapir. Each stage is probably recorded in some fashion in the stratigraphic record. The fountain-fed surges must have emplaced at least the lower part of the coarse-grained facies of the proximal Phase V beds. And slack surges most likely emplaced the fine-grained facies in both the proximal and distal areas. Back-draft deposits, if they exist, must lie at the top of the coarse-grained proximal facies, either as some minor bed forms not yet recognized or as the stratified beds observed in some sections.
The most striking feature of the component data is the overall decrease in the proportion of andesite in the preclimactic eruptions (fig. 27). Tephra erupted during the first June 12 eruption was almost entirely andesite, while the proportion of dacite increased in the next two vertical eruptions. Dacite proportions increased irregularly through the surge events; the uppermost bed analyzed at site 11, the most proximal site, was devoid of andesite. This decrease helps correlate the preclimactic pyroclastic-flow deposits in the Maraunot River valley (site 24) with specific eruptions, as shown in figure 27. As indicated by the absence of dacite pyroclasts, the first (lowest) pyroclastic-flow deposit was almost certainly produced during the first vertical eruption (0851 on June 12). We know from photographs that the second pyroclastic flow was emplaced before the June 14 vertical eruption. The component data therefore indicate that this second flow was emplaced either during the second vertical eruption of June 12 or, more likely, during the June 13 eruption.
The component data show that the vanguard magma--the first magma to reach the surface--was andesite. This was progressively replaced by dacite magma. The dearth of andesite pyroclasts in the deposits of the last few preclimactic eruptions suggests that the climactic eruption began shortly after the vanguard magma was fully replaced by dacite magma.
Only one type of dacite pumice is present in the preclimactic deposits. The climactic (Phase VI) deposits, in contrast, contain two types: white, phenocryst-rich pumice, the dominant type, and tan, phenocryst-poor pumice (Pallister and others, this volume). The preclimactic dacite pumice closely resembles the white, phenocryst-rich type found in the climactic deposits.
The most striking feature of the density distributions (fig. 28) is the high proportion of dense lithic clasts in the preclimactic deposits relative to the climactic deposits. The only apparent difference between the preclimactic dacite pumice and dacite lithic clasts is their vesicularity. The same is true of the andesite scoria and andesite lithics. We strongly suspect that this reflects extensive preeruption degassing of magma at the top of the magma column (Hoblitt and Harmon, 1993) during the periods of repose between preclimactic eruptions.
The "vigor" of the preclimactic explosive eruptions apparently decreased from the first vertical eruption of June 12 until the last surge of June 15, shortly before the start of the climactic eruption. This assertion is based on several indirect lines of evidence because we have no single objective measure of vigor. Perhaps the most compelling evidence comes from the strata produced by the eruptions. In both Phase IV and Phase V deposits--the sequence of fall deposits from the vertical eruptions and the sequence of surge deposits from the eruption fountains--the beds tend to thin and become finer grained upwards. This tendency suggests that, within each of the two phases, the quantity of ejecta declined from one eruption to the next, as did the height to which it was ejected. The progression in eruption-plume heights estimated from weather-radar observations (fig. 33) supports the latter suggestion: heights decrease irregularly from >24 km for the June 12 and 13 eruptions (the radar system had a ceiling of 24 km) to less than 10 km just before the climactic eruption.
Eruption-duration estimates based on SSAM records (fig. 34) also suggest waning eruptive vigor. These estimates assume that periods of high-amplitude, low-frequency seismicity correspond to eruptive activity. On this basis, eruption duration dropped precipitously during the first two vertical eruptions and then remained roughly the same or perhaps increased somewhat just before the climactic eruption.
Another notable feature of the preclimactic eruptive sequence is the general decline in the duration of repose between successive eruptions (fig. 35).
In continuous eruptions, the change from convecting vertical columns to pyroclastic fountains (column collapse) is usually rationalized in terms of either an increase in the vent diameter, a decrease in volatile content, or both (Wilson and others, 1980). We concluded above that the magma involved in the preclimactic eruptions had undergone substantial degassing; this would both lower the volatile content and increase the mean pyroclast density. One could argue that degassed magma became progressively more abundant from eruption to eruption until the net volatile content was insufficient to produce a convective eruption plume. If this were true, we would expect mean clast density to increase from eruption to eruption. With data currently available, the evidence for a systematic increase in mean clast density is not compelling (fig. 36). Some surge beds do indeed have greater mean densities than the preceding fall deposits, but others do not. The observed variations could be due to density segregation during transport rather than to initial density distributions; a substantially greater quantity of density data, collected from a broad area, would be necessary to resolve the ambiguity.
Figure 33. Plume height (in kilometers, as determined by weather radar) versus time for the preclimactic eruptions. Triangles indicate plume height exceeded radar observation limit. Line gaps indicate one or more eruptions for which data are unavailable.
Figure 34. Eruption duration (based on SSAM data) versus time for the preclimactic and climactic eruptions. Line gaps indicate one or more eruptions for which data are unavailable.
Figure 35. The duration, in minutes, of the successive preclimactic repose intervals. Bars absent where duration data unavailable.
Figure 36. Mean clast density for two composite stratigraphic sections, both of which employ the same Phase IV and Phase VI fall deposits. Phase V surge deposit density data from (A) site 23 and (B) site 11. Filled squares, mean density of juvenile clasts; open squares, mean density of juvenile and nonjuvenile clasts.
The volatile content may have been increasing, rather than decreasing, as the proportion of dacite increased through the preclimactic eruptions. This is because the andesite magma was a mixture composed of dacite and basalt magmas (Pallister and others, 1992; Pallister and others, this volume), and basalt magmas are generally believed to possess a lower volatile content than dacite magmas. Thus, the undegassed hybrid andesite magma quite likely had a lower volatile content than that of the undegassed dacite magma. Because the proportion of dacite increased through the preclimactic eruptions, the mean volatile content should have been increasing.
An increase in vent diameter seems to be a more viable explanation for the change in eruptive behavior, although this explanation seems inconsistent with a progressive decline in quantity of ejecta.
Perhaps the change in behavior from Phase IV to Phase V was indeed somehow related to an increase in vent diameter or a decrease in the volatile content, although we have here a series of eruptions rather than a single continuous eruption. It is not obvious, however, how a systematic change in either volatile content or vent diameter could produce all of the patterns exhibited by the preclimactic eruptions.
Any model of Mount Pinatubo's eruptive behavior must account for: (1) the occurrence of a series of preclimactic eruptions, (2) the progressive decline in vigor of these eruptions, (3) the decline in the duration of successive repose intervals, and (4) the initial decline in eruption duration, perhaps followed by an increase in duration just before the culminating eruption. We present here a phenomenological model that accounts for these gross behavioral patterns.
Why did Pinatubo produce a series of smaller eruptions instead of building directly to the climactic eruption? Magma obviously rose to the surface with sufficient pressure to begin and sustain brief eruptions. Eruptions ended because magma pressure dropped below some minimum value necessary to sustain an eruption. But, repeatedly, the pressure increased sufficiently to begin new eruptions. Because the pressure dropped when material was ejected during eruptions, it is reasonable to assume that the pressure increase during repose intervals was caused by new magma flowing into the conduit. This implies that the conduit--the plumbing system that delivered magma to the surface--dilated elastically in response to the magma influx during repose intervals. During eruptions the conduit contracted, as stored elastic strain energy forced magma to the surface. The volumes of magma expelled by the preclimactic eruptions are small; the largest (the first June 12 eruption) is equivalent to a sphere with a diameter of about 200 m (Paladio-Melosantos and others, this volume).
Why did the eruptive vigor decline through the preclimactic eruptions? In response to magma influx, the pressure in the conduit rose to the value necessary to overcome the confining pressure at the top of the magma column. Apparently, the confining pressure at the top of the magma column declined from eruption to eruption. So in each eruption, successively less magma was forced to the surface, at successively lower pressure.
A decline in confining pressure after the first eruption is expectable. Before Mount Pinatubo awakened in April of 1991, the bulk of its edifice consisted of a plug dome, a relic of activity about 5 centuries earlier (Newhall and others, this volume). The old dome was presumably connected to the parent magma body by a conduit whose magma had long since cooled and solidified. When new magma began to force its way upward, apparently in response to a magma-mixing event deep within the magma chamber (Pallister and others, this volume), it had to force open a new pathway to the surface. The magma was volatile saturated in the deep magma reservoir (Gerlach and others, this volume), but vesicle growth accelerated as the magma ascended and additional water was exsolved from the melt. When the magma reached the depth at which growing vesicles became interconnected, juvenile gases were able to escape upward through fractures (Eichelberger and Westrich, 1983; Eichelberger and others, 1986; Hoblitt and Harmon, 1993), even though the magma had not yet reached the surface. These gases were probably responsible for the high SO2 values measured in late May (Daag, Tubianosa, and others, this volume). At first the vanguard magma degassed freely. But as degassing continued, vesicles in the magma collapsed and microlites began to form; these increased magma viscosity. The degassed magma then became an impermeable, viscous plug retarding the escape of gases and slowing the magma's upward progress. This explanation is consistent with the decreasing SO2 flux observed in late May and early June (Daag, Tubianosa, and others, this volume). But because undegassed magma continued to flow from depth and to accumulate behind the degassed plug, the pressure behind the viscous plug increased, even though the rate of gas escape dropped. Eventually, on June 7, the magma pressure increased sufficiently to force the degassed vanguard magma to the surface, and an andesite dome began to grow. The pressure beneath the growing dome continued to increase because the rate of extrusion was less than the rate of intrusion beneath the dome. Finally, at 0851 on June 12, the magma pressure increased sufficiently to overcome the confining pressure, and the first vertical eruption began. About 36 min later (table 1), the pressure dropped to a value less than that required to sustain an eruption. The vent choked with pyroclasts, and the eruption ended.
The system then began to repeat the behavior it followed before the first eruption--that is, degassing of the top of the magma column, continuing influx of new magma into the conduit, and consequent pressure increase. However, the system had changed somewhat. The original narrow, tenacious plug of degassed magma had been at least partly replaced with a pyroclast-filled vent. The pressure threshold for initiating the second eruption would be lower than that of the first. But what about subsequent eruptions?
Two processes began at the end of each eruption: repressurization due to magma influx, and degassing at the top of the magma column. Degassing increases viscosity and, so, increases the confining pressure. Gas loss is a diffusive process. For simplicity, we assume here that gas loss from the top of the magma conduit approximates the diffusive loss of gas from a semiinfinite medium whose surface gas concentration is maintained at zero (Crank, 1990, p. 32). That is, we assume that the total gas loss and, therefore, the confining pressure, is proportional to the square root of time elapsed since the end of the previous eruption. At first, confining pressure increases more rapidly than magma pressure in the conduit. But because the rate of gas loss decreases with time, the rate of confining-pressure growth also decreases with time. Eventually, the magma pressure exceeds the confining pressure, and another eruption begins. After the first eruption, the time dependence of gas loss from the top of the magma column probably changed little from eruption to eruption. If the magma-pressurization rate remained constant, the confining pressure would remain constant, and the vigor of successive eruptions would all be about the same, as would the durations of successive eruptions and repose intervals. But if the magma-pressurization rate was increasing with time, progressively less time would be available for gas loss to increase the confining pressure before the magma pressure exceeded the confining pressure. The magma pressure would exceed the confining pressure at successively lower values, eruptions would begin at successively lower magma pressures, and the vigor of successive eruptions would decline. Because the vigor of the preclimactic eruptions successively declined, we assume that the magma-pressurization rate was increasing with time.
The duration of repose intervals declined from eruption to eruption because the magma-pressurization rate was increasing with time. The higher rate reduced the repose interval in two ways. First, it reduced the time necessary for the magma pressure to rise to the confining pressure. Second, it reduced the time available for the confining pressure to increase, so confining pressure was less for each successive eruption.
Eruptions persist as long as magma is delivered to the surface at a pressure that exceeds some minimum threshold value. Ejection of magma in an eruption causes pressure to drop, but this drop is offset by the influx of new magma. If the pressurization rate is increasing with time, as we have suggested, successive eruption durations should increase. But eruption duration also depends on the confining pressure threshold, and these thresholds decreased in successive eruptions. The effect of a decreasing confining-pressure threshold dominates eruption duration at first, but the increasing pressurization rate eventually becomes dominant, and eruption durations lengthen. This dominance becomes absolute when the pressurization rate--the rate at which magma pressure rises as a result of the influx of new magma--exceeds the depressurization rate--the rate at which magma pressure drops as a result of the ejection of magma in an eruption. When this occurs, a sustained climactic eruption begins.
We now have the elements necessary to construct a simple phenomenological model.
P = Pressure at the top of the magma column;
dP/dt = Total rate of change of P;
dPin/dt = Rate of change of P due to influx of new magma;
dPout/dt = Rate of change of P due to ejection of magma during eruptions;
Pclose = Pressure value below which eruption will terminate;
Popen = Confining pressure threshold; that is, the pressure at which an eruption will begin.
dPin/dt is directly proportional to time elapsed since the volcano reawakened;
dPout/dt is constant;
Pclose is constant;
Popen is directly proportional to the square root of time elapsed since the end of the previous eruption.
We increase pressure according to dP/dt = dPin/dt until P = Popen, and then we vary pressure according to dP/dt = dPin/dt -dPout/dt until P = Pclose. Using arbitrary values for the various parameters, we obtain the pressure-variation pattern shown in figure 37. This pattern mimics the gross features of the Pinatubo eruption sequence. The successive repose intervals decrease monotonically, while the successive eruption durations decrease to a minimum and then increase prior to the climactic eruption. The vigor of the eruptions, proportional to the magma pressure, declines before the climactic eruption. The eruptive style would change from vertical convection to pyroclastic fountaining when the ejecta velocity fell below some threshold value necessary for convective rise.
Figure 37. Pressure at the top of the magma column versus time for a phenomenological model designed to mimic Pinatubo's eruptive behavior. The rate of pressurization increases directly with time. The pressure threshold that must be reached before an eruption can start is shown by the line labeled Popen. Once an eruption begins, pressurization is offset by a constant-rate depressurization. As long as the depressurization rate exceeds the pressurization rate, the pressure declines until Pclose is reached. Pclose is the minimum pressure necessary to sustain an eruption. Climactic eruption begins when the rate of pressurization exceeds depressurization rate. Black bars show eruption duration, the spaces between the black bars are repose intervals.
In this scheme, eruptive behavior is controlled primarily by the magma-pressurization rate at the top of the magma column and the rate of gas loss from the top of this column. To produce the pattern of preclimactic eruptions observed at Pinatubo, it is necessary for the magma-pressurization rate to increase with time. Such behavior is probably common for a reawakening volcano, because the magma-pressurization rate is directly related to the magma-supply rate, which is apt to increase with time. Rising magma will erode the conduit margins and increase its mean cross-sectional area and thus increase the flow rate. Flow rate is very sensitive to conduit geometry. In a cylindrical conduit, for example, the flow rate is proportional to the fourth power of the radius (Bird and others, 1960). Clearly, the flow rate increased as Pinatubo went from repose to weeks of accelerating unrest, to days of discontinuous eruptions, to the 9-h climactic eruption. We have restricted our discussion to the period during which the flow rate was increasing with time--the preclimactic period. However, the flow rate must have eventually decreased with time.
The laws that govern eruption patterns are certainly more complex than those we present here. Yet, even the simple assumptions we employ yield patterns that qualitatively mimic Pinatubo's patterns, so our phenomenological model may serve as the foundation for a more physically based model of broader applicability.
Most of the observations and data presented in this paper were acquired with helicopter support provided by the U.S. Air Force, U.S. Navy, U.S. Marine Corps, and the Philippine Air Force. Without exaggeration, this study would not have been possible without that support. Logistical support during the eruptions was provided by the U.S. Air Force and after the eruptions by the Philippine Air Force. Bobbie Myers and B. Arlene Compher kindly assisted with illustration preparation. Roger Denlinger and Dan Dzurisin contributed valuable discussions. Kathy Cashman, Bill Rose, and Chris Newhall reviewed the manuscript and provided numerous insightful comments and suggestions.
Banister, J.R., 1984, Pressure wave generated by the Mount St. Helens eruption: Journal of Geophysical Research, v. 89, no. D3, p. 4895-4904.
Barry, R.G., and Chorley, R.J., 1978, Atmosphere, weather and climate (3d ed.): London, Methuen & Co., Ltd., 432 p.
Bird, R.B., Stewart, W.E., and Lightfoot, E.N., 1960, Transport phenomena: New York, John Wiley & Sons, 780 p.
Crank, John, 1990, The mathematics of diffusion (2d ed.): New York, Oxford University Press, 414 p.
Criswell, C.W., 1987, Chronology and pyroclastic stratigraphy of the May 18, 1980, eruption of Mount St. Helens, Washington: Journal of Geophysical Research, v. 92, no. B10, p. 10237-10266.
Daag, A.S., Tubianosa, B.S., Newhall, C.G., Tuñgol, N.M, Javier, D., Dolan, M.T., Delos Reyes, P.J., Arboleda, R.A., Martinez, M.L., and Regalado, M.T.M., this volume, Monitoring sulfur dioxide emission 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.
Eichelberger, J.C., and Westrich, H.R., 1983, Behavior of water in rhyolitic magma at shallow depth: Eos, Transactions, American Geophysical Union, v. 64, p. 338.
Endo, E.T., and Murray, Tom, 1991, Real-time Seismic Amplitude Measurements (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.
Fisher R.V., 1990, Transport and deposition of a pyroclastic surge across an area of high relief: The 18 May 1980 eruption of Mount St. Helens, Washington: Geological Society of America Bulletin, v. 102, p. 1038-1054.
Folk, R.L., 1980, Petrology of sedimentary rocks: Austin, Hemphills, 184 p.
Foxworthy, B.L., and Hill, Mary, 1982, Volcanic eruptions of 1980 at Mount St. Helens--The first 100 days: U.S. Geological Survey Professional Paper 1249, 125 p.
Gerlach, T.M., Westrich, H.R., and Symonds, R.B., this volume, Preeruption vapor in magma of the climactic Mount Pinatubo eruption: Source of the giant stratospheric sulfur dioxide cloud.
Harlow, D.H., Power, J.A., Laguerta, E.P., Ambubuyog, G., White, R.A., and Hoblitt, R.P., this volume, Precursory seismicity and forecasting of the June 15, 1991, eruption of Mount Pinatubo.
Hoblitt, R.P., 1986, Observations of the eruptions of July 22 and August 7, 1980, at Mount St. Helens, Washington: U.S. Geological Survey Professional Paper 1335, 44 p.
Hoblitt, R.P., and Harmon, R.S., 1993, Bimodal density distribution of cryptodome dacite from the 1980 eruption of Mount St. Helens, Washington: Bulletin of Volcanology, v. 55, p. 421-37.
Hoblitt, R.P., Miller, C.D., Vallance, J.W., 1981, Origin and stratigraphy of the deposit produced by the May 18 directed blast, in Lipman, P.W., and Mullineaux, D.R., eds., The 1980 eruptions of Mount St. Helens, Washington: U.S. Geological Survey Professional Paper 1250, p. 401-419.
Inman, D.L., 1952, Measures for describing the size distribution of sediments: Journal of Sedimentary Petrology, v. 22, no. 3, p. 125-145.
Koyaguchi, Takehiro, and Tokuno, Masami, 1993, Origin of the giant eruption cloud of Pinatubo, June 15, 1991: Journal of Volcanology and Geothermal Research, v. 55, p. 85-96.
Moore, J.G., and Sisson, T.W., 1981, Deposits and effects of the May 18 pyroclastic surge, in Lipman, P.W., and Mullineaux, D.R., eds., The 1980 eruptions of Mount St. Helens, Washington: U.S. Geological Survey Professional Paper 1250, p. 421-438.
Newhall, C.G., Daag, A.S., Delfin, F.G., Jr., Hoblitt, R.P., McGeehin, J., Pallister, J.S., Regalado, M.T.M., Rubin, M., Tamayo, R.A., Jr., Tubianosa, B., and Umbal, J.V., this volume, Eruptive history of Mount Pinatubo.
Oswalt, J.S., Nichols, W., and O'Hara, J.F., this volume, Meteorological observations of the 1991 Mount Pinatubo eruption.
Paladio-Melosantos, M.L., Solidum, R.U., Scott, W.E., Quiambao, R.B., Umbal, J.V., Rodolfo, K.S., Tubianosa, B.S., Delos Reyes, P.J., Alonso, R.A., and Ruelo, H.R., this volume, Tephra falls of the 1991 eruptions of Mount Pinatubo.
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.
Pallister, J.S., Hoblitt, R.P., and Reyes, A.G., 1992, A basalt trigger for the 1991 eruptions of Pinatubo Volcano?: Nature, v. 356, p. 426-428.
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.
Reed, J.W., 1980, Air pressure waves from Mt. St. Helens eruptions: Albuquerque, N. Mex., Sandia National Laboratories, SAND80-1970A, 14 p.
Rogers, J.A., 1989, Frequency-domain detection of seismic signals using a DSP co-processor board, in Lee, W.H.K., ed., IASPEI software volume I: Toolbox for seismic data acquisition, processing, and analysis: International Association of Seismology and Physics of the Earth's Interior, p. 151-172.
Sabit, J.P., Pigtain, R.C., and de la Cruz, E.G., this volume, The west-side story: Observations of the 1991 Mount Pinatubo eruptions from the west.
Sigurdsson, H., Carey, S.N., and Espindola, J.M., 1984, The 1982 eruptions of El Chichón volcano, Mexico: Stratigraphy of pyroclastic deposits: Journal of Volcanology and Geothermal Research, v. 23, p. 11-37.
Scott, W.E., Hoblitt, R.P., Torres, R.C., Self, S, Martinez, M.L., and Nillos, T., Jr., this volume, Pyroclastic flows of the June 15, 1991, climactic eruption of Mount Pinatubo.
Sparks, R.S.J, Moore, J.G., and Rice, C.J., 1986, The initial giant umbrella cloud of the May 18th, 1980, explosive eruption of Mount St. Helens: Journal of Volcanology and Geothermal Research, v. 28, p. 257-274.
Taylor, G.C., 1958, The 1951 eruption of Mount Lamington, Papua: Australian Bureau of Mineral Resources, Geology and Geophysics Bulletin 38, 129 p.
Wilson, Lionel, Sparks, R.S.J., and Walker, G.P.L., 1980, Explosive volcanic eruptions--IV: The control of magma properties and conduit geometry on eruption column behavior: Geophysical Journal of the Royal Astrological Society, v. 63, p. 117-148.
FIRE and MUD Contents
PHIVOLCS | University of Washington Press | U.S.Geological Survey
This page is <https://pubs.usgs.gov/pinatubo/hoblitt2/>
Contact: Chris Newhall
Last updated 06.10.99