Information to Users

Seismic investigations of subsurface volcanic structures and processes at Mount Spurr, Alaska and Soufriere Hills Volcano, Montserrat, West Indies

Item Type Thesis

Authors Power, John A.

Download date 28/09/2021 13:42:49

Link to Item http://hdl.handle.net/11122/9545 INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type o f computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely afreet reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back o f the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zed> Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600

The original manuscript received by UMI contains pages with indistinct print. Pages were microfilmed as received.

This reproduction is the best copy available

UMI

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SEISMIC INVESTIGATIONS OF SUBSURFACE VOLCANIC STRUCTURES AND PROCESSES AT MOUNT SPURR, ALASKA AND SOUFRIERE HELLS VOLCANO, MONTSERRAT, WEST INDIES

A THESIS

Presented to the Faculty

Of the Universitv of Alaska Fairbanks

In Partial Fulfillment of the Requirements

for the degree of

DOCTOR OF PHILOSOPHY

By-

John A. Power. M.S.

Fairbanks. Alaska

December 1998

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9918838

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SEISMIC INVESTIGATIONS OF SUBSURFACE VOLCANIC STRUCTURES AND PROCESSES AT MOUNT SPURR, ALASKA AND SOUFRIERE HILLS VOLCANO, MONTSERRAT, WEST INDIES

John A. Power

RECOMMENDED:

Advisory Committee Chair

Department Head

APPROVED: Dean. College of Science. Engineering and Mathematics

Deap'of the Graduate School

Date

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3

ABSTRACT

Seismological techniques are used to infer the subsurface structures and volcanic processes at two recently active volcanoes: Mount Spurr. Alaska, and Soufriere Hills Volcano. Montserrat. West Indies. The three-dimensional P-wave velocity structure of Mount Spurr is determined to depths of 10 km by tomographic inversion of 3.754 P-wave arrival times from local earthquakes. Results show a prominent low-velocity zone beneath the southeast flank of Crater Peak extending from the surface to 3-4 km below sea level, spatially coincident with an active geothermal system. Beneath Crater Peak an approximately 3-km- wide zone of relatively low velocities correlates with a near vertical band of seismicity, suggestive of a magma conduit. No large low-velocity zone indicative of a magma chamber occurs within the upper 10 km of the crust. In the three years bracketing the 1992 eruptions of Mount Spurr’s Crater Peak vent, approximately 2.500 located events were classified as Volcano-Tectonic (VT) earthquakes. Long-Period (LP) events, or Hybrid events. An unusual mix of VT. LP. and hybrid events at 20 to 40 km depth began coincident with the onset of unrest and peaked shortly after eruptive activity ended. The classified seismic events are combined with geophysical and geological data to develop a simplified model of the magmatic plumbing system of Mount Spurr. The major components of this model are a deep magma source zone at 20 - 40 km depth, a smaller storage zone at about 10 km depth, and a pipe-like conduit that extends to the surface. The frequency-magnitude distribution of earthquakes measured by the 6-value is determined as a function of space beneath Soufriere Hills Volcano, from data recorded between August 1. 1995 and March 31. 1996. A volume of high 6-values (b > 3.0) with a 1.5 km radius is imaged between 0 and 1.5 km beneath English’s Crater and Chance's Peak. This anomaly extends southwest to Gage's Soufriere. At depths greater than 2.5 km. volumes of comparatively low 6-values (6-1) are found beneath St. George's Hill. Windy Hill, and below 2.5 km to the south of English's Crater.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4

Table of Contents

Abstract------3

Table of Contents------4

List of Figures------7

Acknowledgments------9

Chapter 1 Thesis Introduction 1.0 General Introduction------10 1.1 Overview of the Chapters------11 1.2 References------13

Chapter 2 Seismic image of the Mount Spurr Magmatic System 2.0 Abstract------16 2.1 Introduction------16 2.2 Geology and geophysical studies of Mount Spurr------17 2.3 Seismicity of Mount Spurr ------19 2.4 Data and technique------20 2.5 Three-dimensional P-wave V elocity Model for Mount Spurr ------22 2.6 Discussion ------23 2.7 Conclusions------25 2.8 Acknowledgments------25 2.9 References------36

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5

Chapter 3 Seismicity of the Mount Spurr magmatic system during the 1992 Crater Peak eruption sequence 3.0 Abstract------40 3.1 Introduction------41 3*2 Review of the Mount Spurr magmatic system ------42 3*2.1 Eruptive History and Geology ------42 3*2.2 Seismicity and Geophysics ------45 3.3 Data and earthquake classification------46 3*3.1 Seismic instrumentation and recording------46 3.3*2 Classification method ------47 3.4 Results and Interpretation------48 3.4.1 Magmatic system geometry ------48 3.4.2 Seismic activation and evolution------50 3.4*3 Deep seismicity and eruption termination ------52 3.4.4 Eruption forecasting and volcanic hazards------54 3.5 Conclusions ------55 3.6 Acknowledgments------55 3.7 References------68

Chapter 4 Spatial variations in the frequency-magnitude distribution of earthquakes at Soufriere Hills Volcano, Montserrat, West Indies 4.0 Abstract------75 4.1 Introduction------75 4.2 Data------76 4.3 Method------77 4.4 Results------78 4.5 Discussion and conclusions------79

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6

4.6 Acknowledgments------80 4.7 References------85

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5

List of Figures

Chapter 2 2.1 Location map of Mount Spurr and study area* ————* ------26 2.2 Geology and geography of Mount Spurr and surrounding area------27 2.3 Map of geophysical anomalies observed near Crater Peak------28 2.4 Earthquake epicenters near Mount Spurr between 1991 and 1993------29 2.5 One-dimensional velocity model for Mount Spurr area ------30 2.6 Checkerboard reconstruction of Spurr velocity structure ------31 2.7 Plan view of P-wave velocities at selected depths------32 2.8 Cross-sections showing P-wave velocity structure at Mount Spurr------33 2.9 Relocated earthquakes at Mount Spurr------34 2.10 Sketch of the Spurr magmatic system------35

Chapter 3 3.1 Location map for Mount Spurr------56 . 3.2 Range of silica observed at Cook Inlet Volcanoes------57 3.3 Olivine - Quartz - Plagioclase - Diopside tetrahedron for Cook Inlet------58 3.4 Histogram of located earthquakes at Mount Spurr. 1991 - 1993------59 3.5 Map of seismic stations at Mount Spurr------60 3.6 Waveforms of representative seismic events------61 3.7 Normalized velocity spectra of LP. VT, and hybrid seismic events------62 3.8 Epicenter map and cross-section of located earthquakes near Mount Spurr 63 3.9 Focal depth versus time for all located events at Mount Spurr. 1991 - 1993------64 3.10 Magnitudes and cumulative seismic energy release at Mount Spurr, 1991 - 1993- 65 3.11 Epicenter maps and cross-sections for shallow swarms at Crater Peak------66 3.12 Sketch of the inferred Crater Peak magmatic plumbing system------67

Chapter 4 4.1 Montserrat seismic network, July 1995 to March 1996------82

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8

4.2 Maps and cross-sections showing b-values and earthquakes used in this study 83 4.3 Frequency-magnitude distributions from English’s Crater and St. George’s Hill—84

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9

Acknowledgments I would likely have never begun this project without the initial encouragement and support from Max Wyss and Kaye Shedlock. I also wish to express my gratitude to committee members Harley Benz. Doug Christiansen. John Eichelberger. David Stone, and Chairman Max Wyss. who provided advice and reviews of the various manuscripts throughout their preparation. I also wish to thank the co-authors of the various manuscripts prepared for this thesis: Antonio Villasenor who provided the force and drive required to deal with the complex computer codes needed for the tomographc inversion of the Spurr data: Harley Benz for insight into the inversion process and assistance in completing bringing *‘A seismic image of Mount Spurr" (Chapter 2) to rapid publication; Art Jolly for his general assistance and knowledge of Spurr seismicity; Chris Nye and Michele Harbin for providing the geochemical perspective on the 1992 eruption sequence used in Chapter 3: Max Wyss who originally envisioned mapping the frequencv-magnitude distribution at Montserrat; and Joan Latchman who provided local knowledge of seismicity and magnitudes in the eastern Caribbean. I extend special thanks to the many friends and colleagues who helped in one way or another through out my "adventure in geophysics".

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10

Chapter 1 1.0 General Introduction Volcanic earthquakes have long been used to infer the location and size of magma chambers, the processes associated with magma transport, and the velocity of magma ascent (e.g. Koyanagi et al.. 1976: Malone 1983: Klein et ai.. 1987). More recent studies of wave propagation through volcanic areas have been used to map the geologic structures beneath volcanoes (e.g. Benzet al.. 1996: Patane et al.. 1994). Additionally, close seismic monitoring of a number of active volcanoes (e.g. Okada et al., 1981; Malone 1983: Klein et al., 1987; Power et al., 1994: Harlow et al.. 1996). as well as modeling of source mechanisms (e.g. Chouet. 1992; Julian. 1994). has greatly improved our understanding of volcanic processes associated with the various types of seismic events observed at active volcanoes. These advances provide a clearer understanding of subsurface volcanic structures and the processes which lead to magma ascent and eruption. Our understanding of these processes and their seismic expression has provided the basis for a number of successful forecasts of volcanic eruptions. These forecasts were based on changes in the rate of occurrence of volcanic earthquakes, increases in the total seismic energy release, and changes in the character of earthquake waveforms. Forecasts of volcanic eruptions and mitigation of volcanic hazards has proven to be most effective when monitoring of seismic activity is combined with related observations of ground deformation, thermal and potential field measurements, geochemistry, and geological observations of past eruptions, as well as a sound understanding of the individual magmatic system (Swanson et al.. 1985: Tilling 1989). This thesis uses three recently developed seismologic techniques to infer the subsurface structure and volcanic processes at two active volcanoes: Mount Spurr. Alaska, which erupted in 1992: and South Soufriere Hills volcano on Montserrat. West Indies, which at the time of this writing (May 1998) continues to erupt. The studies included in this thesis were selected to improve our understanding of the plumbing of magmatic systems, volcanic processes, and consequently our ability to forecast eruptions at these two hazardous volcanoes.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11

1.1 Overview of the Chapters

This thesis is devided into four chapters: this introduction, and three chapters, which have been prepared as separate journal manuscripts. Each of these manuscripts has its own introduction, discussion of data and analysis, and conclusions, which are specific to the individual chapter. Figures are arranged at the end of each chapter. The presentation of the individual chapters does not reflect the order in which the work was completed. The first manuscript (Chapter 2) uses local earthquake P-wave travel-time tomography to determine the three-dimensional velocity structure of the Mount Spurr area to depths of 10 km. The primary objective of this study is to identify the components of the Mount Spurr/Crater Peak magmatic system which can be imaged based on seismic velocity contrast. Results show a prominent low-velocity zone extending from the surface to 3-4 km below sea level beneath the southeast flank of Crater Peak, spatially coincident with a geothermal system. P-wave velocities in this low-velocity zone are approximately 20% slower than those in the shallow crystalline basement rocks. Beneath Crater Peak an approximately 3-km-wide zone of relative low velocities correlates with a near vertical band of seismicity, suggestive of a magmatic conduit. No large low-velocity zone indicative of a magma chamber occurs within the upper 10 km of the crust. These observations are consistent with petrologic and geochemical studies suggesting that Crater Peak magmas originate in the lower crust or upper mantle and have a short residence time in the shallow crust. This work has been accepted for publication in the Bulletin of Volcanology (Power et al.. 1998). In the next chapter (Chapter 3). we classify all located seismic events within approximately 18 km of Mount Spurr between January 1991 and December 1993 based on our understanding of source processes of the various types of seismic events. The classification system used identifies events as Volcano-Tectonic (VT) earthquakes which reflect the brittle failure of rock. Long-Period (LP) events which are thought to be generated by the movement of magma, and hybrid events which share the characteristics and perhaps the source processes of both VT and LP events. Stacked velocity spectra are used to eliminate wave propagation effects such as scattering, diffraction, and absorption in the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12

observed seismic waveforms. Significant seismic features of the eruption sequence include: ( 1) a distinct swarm of VT earthquakes in August 1991 directly beneath the Crater Peak vent: (2) a caidera - wide increase in VT earthquakes lasting seven months which preceded the June 27 eruption: (3) two shallow swarms of VT earthquakes which occurred on June 5 and June 27. the latter immediately preceding the June 27 eruption; (4) a mix of VT. LP. and Hybrid events at depths of 20 - 40 km which coincided with the onset of seismic unrest and reached a peak after eruptive activity ended. (5) a strong swarm of VT earthquakes which began as the September 16-17 eruption was ending; (6) a prominent swarm of VT earthquakes on November 9-10 at depths of - I to 4 km beneath Crater Peak: and (7) a smaller swarm of VT earthquakes in late December 1992 which locate between 7 and 10 km depth. This chapter has been submitted to the Journal of Geophysical Research. Chapter 4 is a study of the frequency-magnitude distribution of earthquakes measured by the 6-value as a function of space around South Soufriere Hills Volcano. Montserrat. West Indies. We use earthquake hvpocenters and magnitudes calculated by the Montserrat Volcano Observatory between August 1995 and March 31. 1996. The b-value is estimated at the nodal points of a two-dimensional grid using the N nearest earthquakes. Wiemer and W'yss (1997) and Wyss et al.. (1997) describe this relatively new technique. Results show a volume of anomalously high 6-values (6 > 3.0) with a 1.5 km radius at depths of 0 and 1.5 km beneath English's Crater and Chance's Peak. This high b- value anomaly extends southwest to Gage's Soufriere. At depths greater than 2.5 km. volumes of comparatively low 6-values (6-1) are found beneath St. George's Hill. Windy Hill, and below 2.5 km depth and to the south of English’s Crater. Chapter 4 has been accepted for publication in Geophysical Research Letters (Power et al., in press).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13

1.2 References

Benz H.M.. B.A. Chouet. P.B. Dawson, J.C. Lahr. R.A. Page, and J.A. Hole, Three dimensional P and S wave velocity structure of Redoubt Volcano. Alaska. J. Geophys. Res. 101: 8111-8128, 1996.

Chouet, B.A.. A seismic model for the source of long-period events and harmonic tremor, in Volcanic Seismology. LAVCEI Proc. in Volcanology, edited by P. Gasparini, R. Scarpa, and K. Aki. Springer Verlag, Berlin. 3. 133-156, 1992.

Harlow. D.H.. J.A. Power. E.P. Laguerta. G. Ambubuyog, R.A. White, and R.P. Hoblitt. Precursory seismicity and forecasting of the June 15. 1991 eruption of Mount Pinatubo. in Fire and Mud, eruptions and lahars of Mount Pinatubo. Philippines, edited by C.G. Newhall and R.S. Punongbayan. University of Washington Press, Seattle, WA. 285-306, 1996.

Julian, B.R.. Volcanic tremor: Nonlinear excitation by fluid flow. J. Geophys. Res.. 99: 11859-11877. 1994.

Klein. F.W.. R.Y. Koyanagi. J.S. Nakata. and W.R. Tanigawa. The seismicity of Kilauea’s magma system, in Volcamsm in Hawaii, edited by R.W. Decker. T.L. Wright, and P.H. Stauffer, U.S. Geol. Surv.. Prof. Pap., 1350. I0 I9 -1 186. 1987.

Koyanagi. R.Y.. J.D. Unger. E.T. Endo. and A.T. Okamura. Shallow earthquakes associated with inflation episodes at the summit of Kilauea Volcano. Hawaii. Bulletin of Volcanologique. 39: 621-631. 1976.

Malone, S.D.. Volcanic earthquakes: Examples from Mount St. Helens, in Earthquakes: Observations. Theroy. and Interpretation, edited by H. Kanamori and E. Boschi. Elsevier/North Holland. Amsterdam. 436-455. 1983.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14

Okada. H., H. Watanabe, I. Yamahita, and I. Yokoyama. Seismological significance of the 1977 - 1978 eruptions and magma intrusion process of Usu Volcano. Hokkaido, J. Vole, and Geotherm. Res.. 9, 311-344. 1981.

Patane'. D.. F. Fermcci, and S. Greta. Spectral features of microearthqukaes in volcanic areas: attenuation in the crust and amplitude response of the site at Mt. Etna. Italy. Bulletin of the Seismological Society of America. 84. 1842-1860. 1994.

Power. J.A.. J.C. Lahr. R.A. Page. B.A. Chouet. C.D. Stephens, D.H. Harlow. T.L. Murray, and J.N. Davies. Seismic evolution of the 1989 - 1990 eruption of Redoubt Volcano. Alaska. J. Vole, and Geotherm. Res.. 62. 69-94. 1994.

Power. J.A.. A. Villasenor, and H.M. Benz. A seismic image of Mount Spurr magmatic system. Bulletin of Volcanology, 60. 27-37. 1998.

Power. J.A.. M. Wyss. and J.L. Latchman. Spatial variations in the frequency-magnitude distribution of earthquakes at Soufriere Hills Volcano, Montserrat. West Indies. Geophy. Res. Let., in press.

Swanson. D.A.. T.J. Casadeval. D. Dzurisin. S.D. Malone. C.G. Newhall. and C.S. Weaver. Forecasts and prediction of eruptive activity at Mount St. Helens. L.S.A.: 1975-1984. J. Geodynamics. 3. 397-423. 1985.

Tilling, R.I.. Volcanic Hazards and their mitigation: progress and problems. Rev. Geophys.. 27. 237-269. 1989.

Wiemer. S.. and M. Wyss. Mapping the frequency-magnitude distribution in asperities: an improved technique to calculate recurrence times? J. Geophys. Res.. 102. 15115-15128. 1997.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15

Wyss. M., K. Shimazaki. and S. Wiemer. Mapping active magma chambers by 6-values beneath off-Ito Volcano. Japan. J. Geophys. Res.. 102. 20413-20433. 1997.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16

CHAPTER 2 Seismic Image of the Mount Spurr Magmatic System1

2.0 Abstract The three-dimensional P-wave velocity structure of Mount Spurr is determined to depths of 10 km by tomographic inversion of 3.754 first arriving P-wave times from local earthquakes recorded by a permanent network of 11 seismographs. Results show a prominent low velocity zone extending from the surface to 3-4 km below sea level beneath the southeast flank of Crater Peak, spatially coincident with a geothermal system. P-wave velocities in this low-velocity zone are approximately 20% slower than those in the shallow crystalline basement rocks. Beneath Crater Peak an approximately 3-km-wide zone of relative low- veiocities correlates with a near vertical band of seismicity, suggestive of a magmatic conduit. No large low-velocity zone indicative of a magma chamber occurs within the upper 10 km of the crust. These observations are consistent with petrologic and geochemical studies suggesting that Crater Peak magmas originate in the lower crust or upper mantle and have a short residence time in the shallow crust. Earthquakes relocated using the three-dimensional velocity structure correlate well with surface geology and other geophysical observations, and thus they provide additional constraints on the kinematics of the Mount Spurr magmatic system.

2.1 Introduction Mount Spurr is an active, composite volcano situated at the northeastern end of the Aleutian arc approximately 125 km west of Anchorage. Alaska (Fig. 2.1). This predominantly andesitic volcano has erupted twicein historic time, a single eruptive event in 1953 (Juehle

1 Published as: Power JA. Villasenor A. Benz HM. (1998) Seismic image of the Mount Spurr magmatic system.

Bull ofVolcanal 60:27-37.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17

and Coulter 1955) and a sequence of three eruptive events on 27 June. 18 August, and 16-17 September 1992 (Alaska Volcano Observatory Staff 1993). Both historic eruptions were from the Crater Peak vent, a small composite cone on the south flank of Mount Spurr (Fig. 2.2). Each of the 1992 eruptions lasted about four hours and produced between 12 and 15 million cubic meters D.R.E. (Dense Rock Equivalent) of andesitic magma (53 to 57% SiO:). The total 1992 erupted volume was approximately 4% of the output from the May 18. 1980. Mount St. Helens eruption. The 1992 Mount Spurr eruptions were preceded by 10 months of increasing seismic activity. Following the 1992 eruption sequence elevated seismic activity was observed into 1994. Seismic activity associated with the 1992 eruption was monitored by a network of 7 to 11 radio-telemetered short-period seismic stations operated by the Alaska Volcano Observatory (Fig. 2.2). In this paper, first arriving P-wave phases recorded by the local seismic network between I January 1991 and 31 December 1993 are used to determine a three-dimensional P- wave velocity structure of Mount Spurr to a depth of approximately 10 km. We describe the available seismic data and technical aspects of the tomography and ray-tracing techniques used to determine the three-dimensional velocity structure. We then discuss the results of the tomographic inversion, place them in context with the known geology and results of earlier studies, and discuss the implications of the results for our understanding of the Spurr magmatic svstem and associated volcanic hazards.

23. Geology and Geophysical Studies of Mount Spurr Nye and Turner < 1990) estimate that volcanism at Mount Spurr began about 250.000 years ago. and formed ancestral Mount Spurr. composed of interbedded flows of andesite and more mafic intrusions. The Mount Spurr volcanic sequence overlies intrusive crystalline basement composed of granite and quartz diorite (Reed and Lanphere 1969) that crops out west of Mount Spurr (Fig. 2.2). East of Mount Spurr is a Tertiary conglomerate, sandstone, siltstone. and ash-flow sequence that is faulted against the granitic basement rocks. Basement rocks south of the Chakachatna River are composed of granodiorite. Approximately 15.000 years ago ancestral Mount Spurr had a large sector failure, producing a Bezymianny-style collapse caldera approximately 4.5 km wide (Nye and Turner 1990). Following the collapse two vents formed. Mount Spurr summit in the center of the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18

caldera and Crater Peak in the caldera breach (Fig. 2.2). There is no evidence of both vents existing prior to collapse (C.J. Nye. pers. comm.). Spurr summit vent was active between 8.000 and 5.000 years ago. erupting andesitic magma containing 60 - 639£ SiO?, while the Crater Peak vent erupted more mafic magma (53 - 57 *7c SiO;). Working with tephra layers, Riehle (1985) finds evidence for 35 eruptions from the Crater Peak vent over the last 6.000 years. This provides a minimum indication of the long-term activity of Crater Peak, as not all tephra layers have been preserved. The oldest Crater Peak tephras predate the youngest Mount Spurr ash layers and indicate that the two vents were active simultaneously (Riehle 1985). Magma from the 1992 eruptive sequences is higher in silica yet contains lower concentrations of some incompatible elements (e.g.. Rb) compared to the 1953 magma. Based on these differences. Nye et al. (1995) suggest that the recent magma was generated in the lower crust or upper mantle, separately from the 1953 magma. The 1992 eruptions produced two similar types of andesite that differ only in the composition of the groundmass glass. The observed differences between the two andesites were attributed to progressive crystallization or zonation of a single magma body that was tapped for all three 1992 eruptions (Harbin et al. 1995: Neal et al. 1995). Harbin et al. (1995) find pristine hornblende crystals in the 1992 magma, so that the magma apparently did not spend significant time in the upper crust. In 1985. a program of geothermal exploration using self-potential, controlled-source audiomagnetotelluric soundings and helium and mercury soil analyses was conducted between Crater Peak and the Chakachatna River (Turner and Wescott 1986: Wescott et al. 1988: Moore 1990). This exploration was motivated by the presence of active fumaroles on Crater Peak, as well as the presence of hot springs and the proximity of Mount Spurr to communities in south-central Alaska. These studies revealed prominent anomalies in resistivity, self-potential, and elevated mercury and helium concentrations on the southeast flank of Crater Peak. Based on the juxtaposition of these different anomalies, as well as the presence of hot springs. Turner and Wescott (1986) suggested that a geothermal system exists beneath the southeast flank of Crater Peak. Figure 2.3 indicates the location of the potential fields and geochemical anomalies previously identified as possible target zones for geothermal production (Turner and Wescott 1986). The geometry of these anomalies, and the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19

local geology, suggest that the Crater Peak geothermal system may be contained in a south- dipping sandstone or porous tephra layer that is capped at the top by lava flows from Crater Peak at roughly 1 km above sea level. The resistivity anomaly extends to depths of approximately 200 m above sea level, where modeling suggests it overlies the granitic basement rocks (Turner and Wescott 1986).

23 Seismicity of Mount Spurr Using earthquake data collected in the Mount Spurr region between 1981 and 1991. Jolly et al. (1994) observed a regional shoaling of earthquake hypocenters beneath the Spurr volcanic cone. They attributed the shoaling to the presence of warm ductile crust beneath the volcano, which would not fail in a brittle manner given usual strain rates. Axes of earthquake focal mechanisms indicated that the principal regional stress direction was oriented N30°W, concordant with convergence of the Pacific and North American plates, while directly beneath the volcano the principal stress direction was oriented almost vertically. Jolly et al. (1994) suggest that the deviation in principal stress directions beneath Mount Spurr might result from either the weight of the Spurr volcanic pile or doming caused by magmatic processes. The 1992 eruptions were preceded by 10 months of increased earthquake activity beneath Crater Peak, the summit of Mount Spurr. and under the north caldera rim. Most of these precursory earthquakes were located between -3.2 and 10 km deep (negative depths refer to height above sea level), although a few events were observed as deep as 30 km. Following the June 27 eruption, the volcano entered a period of relative seismic quiescence at shallow depths that lasted until the August 18 eruption. Thereafter, the seismicity increased and reached a maximum shortly after the September 16-17 eruption. The number of locatable seismic events then began to decline. Two notable swarms of earthquakes occurred in November and December 1992. but seismicity continued to decline through 1993. Earthquake locations and waveforms from this period are described by Power et al. (1995): McNutt et al. (1995) described the observed episodes of volcanic tremor and eruption seismicity. Based on the space-time pattern of earthquake hypocenters between January 1991 and December 1993. Power et al. (1995) suggested that the 1992 precursory seismicity resulted

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20

from the intrusion of magma at depths of 5 to 15 km. The November and December 1992 swarms were interpreted as intrusions of magma that did not reach the surface. An alignment of deeper earthquake hypocenters dipping southeastward from Crater Peak to roughly 40 km depth was found to be a mix of volcano-tectonic earthquakes and long-period events (Power and Jolly 1994). attributed to magma transport and stress adjustments associated with the 1992 eruptions (Power et al. 1995). Wiemer and McNutt (1997) calculated the spatial distribution of b -values beneath Crater Peak to depths of 12 km using earthquake hypocenters which occurred between 1991 and 1995. A prominent zone of high ^-values between 2.3 and 4.5 km depth was attributed to alteration of rock caused by the vesiculation of ascending magma. They also suggested that higherb -values between 10 and 12 km Kb - 1.2) and between 20 and 40 km depth (b ~ 1.6) might be the result of magma reservoirs.

2.4 Data and Technique Between January 1991 and December 1993. over 2,500 earthquakes were recorded in the Mount Spurr area (Jolly et al. 1996) by a network of 7 to 11 seismic stations within 20 km of the volcano's summit (Fig. 2.4). Analog waveform data from these stations were telemetered to the Alaska Volcano Observatory in Fairbanks and recorded on a digital event-detection recording system. For this study, we selected a subset of 525 well-recorded earthquakes that had at least six P arrival-times and an azimuthal gap of less than 180°. The earthquakes were declustered by not allowing any two hypocenters to lie within 500 m of each other. Thus the earthquakes selected for the tomographic inversion were well distributed throughout the model volume. The tomographic inversion method used in this study is described in detail by Benz et al. (1996) and has been applied to a number of other volcanoes such as Redoubt Volcano. Alaska (Benz et al. 1996). Mauna Loa and Kilauea. Hawaii (Okubo et al. 1997), and Clear Lake. California (Stanley et al. in press). In this technique, local earthquake arrival times are inverted to determine simultaneously both the three-dimensional P-wave velocity structure and the earthquake locations. Theoretical travel times are calculated using the finite difference technique of Podvin and Lecomte (1991). which is well suited for computing travel times in media with large lateral velocity perturbations. The area surrounding the volcano is

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21

parameterized as a grid of constant velocity cells. Perturbations to the slowness (inverse of velocity) are computed using an LSQR solver (Paige and Sanders 1982) with smoothing constraints applied to control the degree of model roughness. A 20 x 20 km area centered on Mount Spurr was selected for this study. The top of the model was 4 km above sea level and the base at 40 km below sea level. This volume was parameterized using uniform 1 x I x I km constant-velocity cells. A finer 0.25 x 0.25 x 0.25 km grid was used for travel-time calculations. Cell dimensions for both travel-time calculation and velocity inversion were determined empirically through a series of tests that addressed the effects of varying cell sizes and smoothing values. Varying the weight that controls the smoothing constraint from 50 to 10 over approximately 15 iterations results in a smooth velocity model that significantly reduces the travel-time RMS. The starting velocity model for our three-dimensional inversion (Fig. 2.5) is a smoothed version of the one-dimensional model of Jolly et al. (1994). which was obtained by forward trial-and-error modeling of a subset of earthquakes recorded beneath Mount Spurr and is used for routine earthquake location. Because the LSQR solver used does not allow for the computation of model resolution and covariance, we used empirical reconstruction of synthetic anomalies to estimate model resolution. Figure 2.6 shows a comparison of a synthetic P-wave checkerboard velocity model and its reconstruction using the method of Benz et al (1996). Synthetic P-arrival times were computed using the source-receiver geometry of the observed data and the checkerboard velocity model (Fig. 2.6). High and low velocity anomalies are ±10% relative to a homogeneous model of 5.8 km/s. Each anomaly is 5 x 5 x 5 km in dimension. The checkerboard velocity model is smoothed, so there is a gradational transition between high and low velocities. The initial velocity model used in the synthetic reconstruction is a homogeneous P-wave velocity model with a velocity of 5.8 km/s. Figure 2.6 shows that the shape and amplitude of the checkerboard pattern are well reproduced beneath Mount Spurr to depths of 10 km. These tests also show the effects of smearing along the edges of the model, where ray coverage is sparse.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22

2.5 Three-dimensional P-wave Velocity Model for Mount Spurr A set of 3.754 P-wave arrival times was inverted to determine the three-dimensional structure beneath Mount Spurr. Re-location of the earthquakes in the three-dimensional structure resulted in an arrival-time residual RMS of 0.19 s. a 30% reduction from the initial arrival-time RMS of 0.27 s in the one dimensional starting model. Shown in Figure 2.7 are map views of the P-wave velocity variations from -1 to 10- km-depth. The tomographic results show P-wave velocity variations that range from 5 km/s at shallow depth, between the summit, to roughly 6 km/s at depths of 10 km (Fig. 2.7). The most prominent feature is a low-velocity region that extends from near the surface to 5-km- depth beneath the south flank of Mt. Spurr. The three-dimensional velocity structure beneath Mount Spurr is best illustrated by selected cross-sections through our three-dimensional velocity model (Fig. 2.8). The locations of the individual cross sections are shown in Figure 2.9. Cross-section A-A’ shows the velocity structure along a NNW-SSE line that crosses through Mount Spurr summit and Crater Peak vent (Fig. 2.8). A low-velocity body (defined by the 4.8 km/s velocity contour) extends from the surface to a depth of about 4 km. This low-velocity zone is coincident with the target zone for geothermal production determined from analyses of shallow low resistivity, elevated mercury and helium concentrations, and anomalous self potential measurements (Fig. 2.3). A trough-like depression of the velocity contours is observed to depths greater than 20 km. This feature is coincident with a near vertical zone of seismicity, which together define the conduit that likely fed the 1992 eruption sequence. This is also evident in cross-section B-B’. a roughly west-east cross-section centered on Crater Peak. In cross-section B-B’. the seismicity lies along a narrow, steeply dipping zone from the shallow crust beneath Crater Peak to about 15 km depth. Along profile C-C\ which runs roughly east-west beneath the summit of Mount Spurr. a broad subtle depression in velocities occurs (following the 5.6 km/s velocity contour) that is about 4-km- wide. This is the northernmost expression of the shallow low-velocity zone observed beneath the southeast flank of Crater Peak. Zones of high velocities (about 6.0 km/s) are observed below 6 km along the west and east flanks of the volcano and along the southeast flank of the volcano near the Chakachatna river: they correlate with outcrops of granodiorite basement rocks (Fig. 2.2).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23

The average change in the hypocentral location is 1.7 ion horizontally and +4.1 km vertically, for the 525 earthquakes simultaneously relocated in the three-dimensional velocity model. A map view comparison of the initial and final earthquake locations (top Fig. 2.9) shows a tightening of the earthquakes along a northwest trend across the caldera. In addition, the earthquakes cluster closer to the Crater Peak and Mount Spurr vents. Some earthquakes initially located close to the edge of the station distribution relocate outside of the network. The initial locations of these events were probably poor as a result of station geometry. In cross-section (Fig. 2.9). the earthquake locations in the three-dimensional velocity model clearly define a zone of seismicity dipping southeastward from directly beneath the Crater Peak vent to 20-km depth. We infer that these hypocenters define the width and extent of the conduit through which magma is transported.

2.6 Discussion The velocities found at Spurr are somewhat faster than those typically seen at shallow depths at active arc volcanoes, perhaps a result of active glaciation which has kept the accumulation of thick pyrociastic deposits to a minimum. The velocity variations observed in the three-dimensional velocity model are consistent with the rock types mapped in the area (Nye and Turner 1990). Areas of higher-than-average velocities are spatially coincident with crystalline basement rocks, and areas of slower-than-average velocities correspond to sedimentary units or zones of hydrothermal alteration. The largest velocity anomaly observed in the three-dimensional model is the lower- than-average lobe at depths less than 4 km beneath the southeast flank of Crater Peak (Fig. 2.8). P-wave velocities in this area are between 4.5 and 5.0 km/s. an approximately 20 9c decrease relative to the crystalline basement rocks. The low-velocity zone is spatially consistent with the areas of anomalous resistivity, self potential, and mercury and helium values identified by Turner and Wescott (1986), Westcott et al. (1988) and Moore (1990) (Fig. 2.3). We propose that the low P-wave velocities reflect hydrothermal alteration associated with the geothermal system. No increase in seismicity associated with the 1992 eruptions of Crater Peak was observed in this area (Power et al. 1995). This lack suggests that the cause of the low velocities is not directly linked to the Mount Spurr-Crater Peak magmatic system and supports the geothermal interpretation for this area. No low-velocity

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24

zones are associated with the other areas of low resistivity and high mercury and helium concentrations identified by Turner and Wescott (1986) farther east (Fig. 2.3). However, these anomalies are not as large as those beneath the southeast flank of Crater Peak. Furthermore, these areas of our model are not well imaged due to poor station coverage and few earthquakes. This low velocity zone terminates at depths roughly comparable with the high t-value anomaly identified by Wiemer and McNutt (1997). perhaps defining an area where the hydrothermal system intersects the magma conduit. A generalized sketch showing the subsurface features inferred from results of this study is shown in Figure 2.10. We propose that the broad low velocity zone beneath the summit of Mount Spurr may represent thermally altered or fractured rocks resulting from the long-term egress of magma starting with the ancestral volcano (Fig. 2.10). The velocity contrast is not as great as that under the southeast flank of Crater Peak, it occurs at depths that the synthetic data (Fig. 2.6) indicate can be relatively well imaged given the distribution of sources and receivers used in this study. The velocity contrast beneath the summit of Mount Spurr is much smaller than we would expect if an active magma chamber existed in this area. Hypocenters relocated beneath the summit of Mount Spurr delineate a north-dipping structure 3 - 7 km deep. Almost all the earthquakes in this area occurred during the 10-month precursory period and were attributed to stress adjustments resulting from the influx of magma at 5-15 km depth beneath Crater Peak (Power et al. 1995). The tightly clustered relocated hypocenters suggest structural control for this seismic zone, possibly the reverse fault mapped east of Crater Peak by Nye and Turner < 1990). A third low-velocity zone dips southeastward from the base of Crater Peak to depths of 10-15 km. It is roughly coincident with an alignment of earthquake hypocenters recorded after the onset of eruptive activity in 1992. but these hypocenters extend to depths of 35 to 40 km. These earthquakes were thought to define a conduit structure that fed the 1992 eruptions (Power et al. 1995). We interpret the localized downwarp of the velocity contours beneath Crater Peak vent as a zone of highly fractured and thermally altered rocks suggestive of an upper crusta! pathway through which magma is transported (Fig. 2.10). The nearly vertical alignment of the relocated hypocenters (Fig. 2.8) may indicate the presence of a steeply dipping conduit to the southeast of Crater Peak.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Velocity anomalies attributed to magma chambers and volcanic conduits have been successfully imaged at a number of volcanoes, including Kilauea (Okubo et al. 1997), Mount Pinatubo (Mori et al. 1996), Mount Redoubt (Benz et al. 1996), Mount St. Helens (Lees and Crosson. 1989. Lees. 1992). and Hengill - Grensdalur (Toomey and Foulger. 1989). We find no evidence of a large low-velocity zone indicative of a magma chamber within the upper 10 km of the crust beneath Mount Spurr and Crater Peak. This result supports the interpretations of Nye et al. (1995) and Harbin et al. (1995), who suggested, based on geochemical and petrologic evidence, that 1992 Crater Peak lava originated within the deeper portions of the crust and did not spend significant time at depths of less than 5 km.

2.7 Conclusions We obtained a detailed three-dimensional P-wave velocity structure of Mount Spurr by tomographic inversion of local earthquake arrival times. The results provide an improved understanding of the subsurface components of the Mount Spurr-Crater Peak magmatic system to depths of 10 km. Improved earthquake locations using the three-dimensional velocity structure provide a framework for mapping the subsurface components of the magmatic system. The dominant P-wave velocity anomalies are thought to represent a shallow hydrothermal system between the surface and 4 km depth, as well as altered or fractured zones beneath both Crater Peak and Mount Spurr indicative of magmatic conduits. No large low-velocitv zone indicative of an active magma chamber was found within the upper 10 km of the crust. We interpret these results in terms of a geothermal system beneath the southeast flank of Crater Peak, a fractured or thermally altered area beneath the summit of Mount Spurr. and a magmatic conduit that dips southeastward from Crater Peak.

2.8 Acknowledgments Preparation of this manuscript benefited greatly from comments and discussions with Chris Nye. Kaye Shedlock. Max Wyss. and Art Jolly. Chris Nye generously provided the map shown in Figure 2.2. Formal reviews of the text and figures were provided by Jonathan Lees and Don Swanson. MEC grant EX 95 38438406 supported AV.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26

156*W 154"W 152'W 150*W 148'W 62'N 62’N

61'N

60'N

59'N 59‘N 156‘W 154'W 152‘W 150*W 148‘W

Figure 2.1) Location map of Mount Spurr and study area: Map of south-central Alaska showing the location of Mount Spurr and Crater Peak, the study area, and nearby volcanoes (solid triangles).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

i * I * I I N 15

61 61 20 N v "• v Faults. Runs Runs Contacts Rocks Basement. Dome Fi«iomenta Alluvium A Fana } P»OlO~Clrtltil f t hK _iX£LAfjAJ!gti_ _iX£LAfjAJ!gti_ >« Crater A Coldera 31 Ou Alluvial'Coliuviai Qd Glacial Oopoaiu < 1 1 I Ternary Clasnc E 2 C=D D«bri& Avalaoi.he Ancestral Ml Spurr I Crystalline [. [. . j CralaiI’ Htnit 0 Pyioclasuc Fan | 'j Mt Spun I Av.ia A A | Av.ia I 'j Spun Mt k 11 U O 'W ai . C API'S C oi 152 152' 152' OO'W i : i r mLi^ i ‘ r mLi^ 152' 152' 10'W , _ / ii/ . > . . .■■by. ^ i j < i« i« I >7 152 152 20'W 152' 152' 20'W blgy'l ' ' 'N 'N 15 20 figure 2.2) Geology geogiapliyami of Mount Spurr and surrounding area (Nye and Turner Seismic I‘WO). slulions used slmly this in are shown as solid triangles; station NCG about is km 15 north of station GUI’, outside the map area (see lug. 2 1).

61 61

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28

Figure 2.3) Map of geophysical anomalies observed near Crater Peak: Topographic map of area surrounding Mount Spurr showing generalized locations of anomalous values of helium, mercury, self-potential, and resistivity identified by Turner and Wescott < 1986). Contour interval is 200 m.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29

30 ' A iNCG DEPTHS 4 ? . < -3.2+ _ 5.0+ 20' ’ . Mt. Spuri£ - 10.0 + z 20.0+ • r ■ * Vi^ ., - CraterPe&l*^ 30.0+ MAGNITUDES a BCl 15’. . -9.0+ L.'^'O N = 0.0+ a , a ^ ch‘ m" - 1.0+ L“M . ■-* - 2.0+ - 3.0+ L lj ^ ^ 61“10'N ■ 2 K>r^ - ~

30 ’ 20 ’ 152WW

Figure 2.4) Earthquake epicenters near Mount Spurr between 1991 and 1993: Epicenters of earthquakes in the Mount Spurr area recorded between January 1991 and December 1993. Symbol size indicates magnitude, symbol type reflects focal depth in km. Seismic stations are shown as solid triangles.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Velocity ikm/s) 5 6 7

5 15

CL0) -OQ J

35 Joily et al. ( 1394) interpolated 40

Figure 2.5) One-dimensional velocity model for the Mount Spurr area. Solid line indicates ID velocity model developed for the Spurr area by Jolly et al. (1994). Dotted line reflects initial interpolated model used as a starting model for seismic tomography performed in this study.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31

Synthetic model Reconstruction

6 1 2 5 * 4

61*20*4

61 15*4

15? ' V W J52' ? tr w 152" lOW 152"0CW 151* W W 152* KPN 152* 20W 1SZ" IffW 1ST OffW

N J

+ ■

20 I- J 0 5 10 IS 20 25 Otstanca (km)

10.0 9.0 - 8.0 -7 0 -6.0 - 5.0 - 4.0 - 3.0 - 2.0 - 1.0 0.0 10 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 Velocity perturbation (%)

Figure 2.6) Checkerboard reconstruction of Spurr velocity structure: Comparison of a synthetic P-wave checkerboard velocity model in map view and cross section (left) and its reconstruction (right). Synthetic P-arrival times used in the reconstruction were computed using the source-receiver geometry of the observed earthquake data and the checkerboard anomalies, which are 10% relative to a homogeneous model of 5.8 km/s. Note that the checkerboard pattern is well reproduced beneath Mount Spurr to depths of 10 km.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32

ML S p u r ML Spurr eafctera nm

CMLiefcvnrtj LjM*/

Layer -I • Ok Layer I - 2k

IS2 KTN 152' 2tTW IS2 tOW I52*OOW 151 SOW 152' 3ffW 152 3 TW 152' I (TV# tS2'OOW tS l SffW

CJwluefKunna Ljkv

Layer 3 - JH

ppppm =i^spipppppppp 42 43 44 4.5 46 4 7 4.8 49 5.0 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 6.0 6.1 6.2 6.3 6.4 vp (km/s)

Figure 2.7) Plan view of P-wave velocities at selected depths. Plots show data derived from three-dimensional arrival time tomography. The seismic stations used to monitor activity (triangles), as well as the Spurr Caldera rim, Chakachamna Lake, and the Chakachatna River are shown for reference.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C' 16

■> surnmil M l Spurr ___ D iM iin i ii (k m ) . 6 10 Caldera urn 16 10 20 B’ B’ C 15 10 10 - 6 ? ? 9 peak Craler llll.lillll t: (kill)

6 B 16 HI 20

Northern caldera run 1 0 l b I h-.t.HH <: t^ih) <: h-.t.HH I ..'(I 2 O u O I lines), with intermediatelines), contours 0.2 ofkm/s shown dotted as lines. Figure 2.S) Figure 2.S) Cross sections showing the P-wave structure Mountvelocity al Sptnr. three-dimensional P-wave Cross structure velocity sectionsbeneath showingMount the Spurr. Locations of are profiles shown Open Figure in circles 2.9. denote hypocenters ofearllu|uakes study. used this in Velocity contour 0.4 km/s is (solid interval

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34

Original locations Relocations

'j' vr.v : i1: :i j ■:-.7 j :irw vn

Northern Northern caiaera rim caldera rim Crater Mt. Spurr Peak summit A’ Chakachatna river

V1. ■v-'Sk.c , », w J *** *

•' h‘

*’ri r«snj

Figure 2.9) Relocated earthquakes at Mount Spurr: Plan views and cross sections showing the initial locations of the 525 earthquakes used in this study based on the I-D velocity model of Jolly et al. 11994) and the final locations based on the relocation in the three dimensional P-wave velocity structure. A - A'. B - B‘ and C - C' are locations of cross sections shown in Figure 2.8.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35

Figure 2.10) Sketch of the Spurr magmatic system: Generalized interpretation of the subsurface components of the Mount Spurr magmatic system based on three-dimensional P- wave tomogaphy results. The vertical slice of the cmst is concordant with cross section A-A’ (Fig. 2.8).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36

2.9 References:

Alaska Volcano Observatory Staff (1993) Mount Spurr’s 1992 eruptions. EOS Trans. 74:217- m

Benz HM. Chouet BA. Dawson PB. Lahr JC. Page RA. Hole JA (1996) Three-dimensional P and S wave velocity structure of Redoubt Volcano. Alaska. J Geophys Res 101: 8111-8128

Harbin ML. Swanson SE. Nye CJ. Miller TP (1995). Preliminary petrology and chemistry of proximal eruptive products: 1992 eruptions of Crater Peak. Mount Spurr Volcano. Alaska, in Keith TEC. (editor). The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, USGS Bulletin 2139. 139 - 148

Jolly AD. Page RA, Power JA (1994) Seismicity and stress in the vicinity of Mount Spurr volcano, south central Alaska. J Geophys Res. 99: 15305 - 15318

Jolly AD. Power JA. Stihler SD. Rao LN. Davidson G. Paskievitch J. Estes S. Lahr JC. 11996) Catalog of earthquake hypocenters for Augustine. Redoubt. Iliamna. and Mount Spurr volcanoes. Alaska: January 1. 1991 - December 31. 1993. USGS Open-File Report 96-70

Juehle W. Coulter H. (1955) The Mt. Spurr eruption. July 9. 1953. EOS Trans. 36: 199-202

Lees JM. (1992). The magma system at Mount St. Helens: Non linear high resolution P-wave tomography. J Volcanol Geotherm Res 53: 103-116

Lees JM. Crosson RS. (1989) Tomographic inversion for three-dimensional velocity structure at Mount St. Helens using earthquake data. J Geophys Res 94: 5716 - 5728

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37

McNutt SR. Tytgat G. Power JA (1995) Preliminary analyses of volcanic tremor associated with 1992 eruptions of Crater Peak, Mount Spurr Volcano. Alaska, in Keith TEC. (editor). The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, USGS Bulletin 2139. 161-177

Moore P (1990) An analysis of controlled-source audio-frequency magnetotelluric data from Mount Spurr. Alaska. University of Alaska. Fairbanks. M.Sc. thesis: 142 p.

Mori J. Ebberhardt-Phillips. D.. Harlow. D.H. (1996) Three-dimensional velocity structure at Mount Pinatubo: resolving magma bodies and earthquake hypocenters. in Newhall CG. and Punoungbayan RS, (editors). Fire and Mud, eruptions and Lahars of Mount Pinatubo. Philippines. University of Washington Press. Seattle: 371-382

Neal CA. McGimsey RG, Gardner CA, Harbin ML. Nye CJ (1995) Tephra-fall deposits from the 1992 eruptions of Crater Peak. Mount Spurr Volcano. Alaska: a preliminary report on distribution, stratigraphy, and composition, in Keith TEC. (editor). The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska. USGS Bulletin 2139. 65-81

Nye CJ. Turner DL (1990) Petrology, geochemistry, and age of the Spurr volcanic complex, eastern Aleutian arc. Bull Volcanol 52: 205-226

Nye CJ. Harbin ML. Miller TP. Swanson SE. Neal CA (1995) Whole rock major- and trace- element chemistry of 1992 ejecta from Crater Peak. Mount Spurr Volcano. Alaska, in Keith TEC, (editor). The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska. USGS Bulletin 2139. 119-128

Okubo PG. Benz HM. Chouet B (1997) Imaging the crustal magma source beneath Mauna Loa and Kilauea volcanoes. Hawaii. Geology: 25867 - 25870

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38

Paige CC. Sanders MA. 1982. LSQR: an algorithm for sparse linear equations and sparse least squares. ACM TOMS. 8: 43-71

Podvin R. Lecomte I (1991) Finite difference computation of travel times in very contrasted velocity models: A massively parallel approach and its associated tools. Geophys J Int 105: 271 -284

Power JA. Jolly AD. Page RA. McNutt SR (1995) Seismicity and forecasting of the 1992 eruptions of Crater Peak vent. Mount Spurr volcano. Alaska: an overview, in Keith TEC. (editor). The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska. USGS Bulletin 2139. 149- 159

Power JA. Jolly AD (1994) Seismicity between 10 and 40 km depth at Mount Spurr. EOS Trans. 75. 715

Reed BL. Lanphere MA (1969) Age and chemistry of Mesozoic and Tertiary plutonic rocks in South central Alaska. Geol Soc Am Bull 80: 23 - 43

Riehle JR (1985) A reconnaissance of the major Holocene tephra deposits in the upper Cook Inlet region. Alaska. J. Volcan Geotherm Res 26: 37-74

Stanley D. Benz HM. Walters MA. Villasenor A. Rodriguez B (in press) Tectonic controls on magmatism and geothermal resources in the Geysers-Clear Lake region. California: Integration of new geologic, earthquake tomography, seismicity, gravity, and magnetotelluric data. GSA Bull

Toomy DR. Folger GR (1989) Tomographic inversion of local earthquake data from the Hengill-Grensdalur central voicano complex. Iceland. J Geophys Res 94:17497 - 17510

Turner DL. Wescott EM (1986) Summary and conclusions of the Mount Spurr Alaska geothermal assessment project. In Turner DL and Wescott EM (editors). Geothermal energy

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39

resource investigations at Mount Spurr Alaska. University of Alaska Geophysical Institute Report UAG R-308. 6.1 - 6.6

Wescott EM. Turner DL. Nye CJ. Motyka RJ. Moore P (1988) Exploration for geothermal energy resources at Mount Spurr. Alaska. Geotherm Resources Council Trans 12:203-210

Wiemer S. McNutt. SR. ( 1997)Variations in the frequency-magnitude distribution with depth in two volcanic areas: Mount St. Helens. Washington, and Mt. Spurr. Alaska. Geophys Res Let 24: 189-192

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40

CHAPTER 3

Seismicity of the Mount Spun* Magmatic System During the 1992 Crater Peak Eruption Sequence2

3.0 Abstract In the 3 years bracketing the 1992 eruptions of Mount Spurr's Crater Peak vent, approximately 2.500 earthquakes were detected and located using a local network of as many as 11 seismic stations. Normalized velocity spectra are used to classify located seismic events as Volcano-Tectonic (VT) earthquakes. Long-Period (LP) events, or hybrid events. Analysis of the space-time seismicity pattern produced by these event types provides a framework for understanding the eruptive sequence. Significant seismic features of the eruption sequence include: (I) a distinct swarm of VT earthquakes in August 1991 directly beneath the Crater Peak vent. (2) a caldera-wide increase in VT earthquakes. lasting 7 months, that preceded the June 27 eruption. (3) two shallow swarms of VT earthquakes that occurred on June 5 and June 27, the latter immediately preceding the June 27 eruption. (4) a mix of VT. LP. and hybrid events at depths of 20 - 40 km which began coincident with the onset of seismic unrest and reached a peak after eruptive activity ended. (5) a strong swarm of VT earthquakes that began as the September 16-17 eruption was ending, (6) a prominent swarm of VT earthquakes on November 9-10 at depths of -1 to 4 km beneath Crater Peak, and (7) a smaller swarm of VT earthquakes in late December 1992 that were located between 7- and 10-km depth. The seismicity, combined with existing geophysical and geological data, provides constraints for a simplified model of the magmatic plumbing system of the Crater Peak vent. The major components of this model are a deep magmatic source zone at depths of 20 - 40 km. a smaller storage zone at about 10-km depth, and a pipe-like conduit system that extends to the surface.

' Submitted to Journal o f Geophysical Research for Publication as: Power. J.A.. A.D.Jolly. CJ. N'ye. and M. L. Harbin. Seismicity of the Mount Spurr magmatic system dunng the 1992 Crater Peak eruption sequence.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41

3.1 Introduction During 1992. Mount Spurr's Crater Peak vent erupted three times: on June 27. August 18, and September 16-17. Each of the eruptions was vulcanian to subplinian in character, lasted 3.5 to 4.0 hours, and produced 12. 14. and 15 x 106 m’ of tephra respectively [Neal et al 1995]. The eruptions produced a wide variety of seismicity, which included volcano-tectonic (VT) earthquakes, long-period (LP) events, hybrid events, periods of volcanic tremor, and eruption signals. The dominant features of this complex seismic sequence include a prominent swarm of VT earthquakes beneath Crater Peak in August 1991. a 7-month-long precursory VT swarm that spanned the Spurr caldera. the eruptions themselves, strong sequences of volcanic tremor, two swarms of VT earthquakes in November and December 1992. and a mix of VT. LP. and hybrid events at depths of 20 to 40 km. Many improvements have recently been made in relating the different types and characteristics of seismic events observed at active volcanoes with the source or causative processes. This has evolved largely by studying data from a number of eruptions at instrumented volcanoes [Okada et al.. 1981: Malone 1983; Klein et al.. 1987; Power et al.. 1994: Harlow et al.. 1996] and theoretical modeling [Aki and Koyanagi. 1981: Chouet. 1992: Julian. 1994: Chouet et al.. 1994; and Chouet 1996]. The advantage of classifying seismic events on the basis of their source characteristics is that the observed seismicity can then be related to volcanic processes. Chouet et al.. [1994] separate volcanic seismicity into two basic processes. The first process includes events caused by the brittle failure of rock as a result of stresses induced by magmatic processes: we refer to these as "VT earthquakes". VT earthquakes generally have impulsive P and S phases and a relatively broad spectrum with energy between 1 and 15 Hz. The second process consists of events in which fluid plays an active role in the generation of seismic waves; we refer to these as "LP events". LP events commonly begin with a low-amplitude high-frequency P phase when recorded at small epicentral distances: this phase is generally followed by a poorly defined S phase and a low- frequency coda that is strongly peaked close to 2 Hz. Events that share the characteristics of both LPs and VTs are referred to as “hybrids” after Lahr et al., [1994], In this paper, we develop a model of the Spurr magmatic system based on a classification of all located seismic events within roughly 20 km of the summit of Mount Spurr between January 1. 1991 and December 31. 1993 and on supporting geologic

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42

observations. We begin with an overview of the geology and seismicity of Mount Spurr with special reference to the 1992 events. A review of the seismic data selected for this study follows with a discussion of the methodology for calculating waveform spectra, which is used to classify all located seismic events as VT, LP, or hybrid. Using our understanding of the source processes of these events, the temporal and spatial distribution of seismic activity, as well as results from earlier geological and geophysical studies, we develop a simplified model of the Crater Peak magmatic system and speculate on the volcanic processes active during the 1992 eruption sequence.

3.2 Review of the Mount Spurr Magmatic System

3.2.1 Eruptive History and Geology: Mount Spurr is located about 125 km west of Anchorage, Alaska (Figure 3.1). Through much of its history. Mount Spurr erupted andesites of fairly uniform composition (58 - 60% SiO:) from the summit vent, which formed a large stratocone. Approximately 15,000 years ago, this ancestral Mount Spurr underwent a large sector failure that produced a Bezymianny- style collapse/landslide caidera [Nye and Turner, 1990]. Following the collapse, two vents formed. The first vent in the center of the newly formed caidera erupted a more silicic andesite (60 - 63% SiO;) between 8.000 and 5.000 years ago. and forms the present summit of Mount Spurr. The second vent. Crater Peak, formed in the caidera breach and erupted more mafic magma (53 - 57% SiO:). The oldest Crater Peak tephras predate the youngest Spurr tephra layers, indicating that for some period of time, the two vents were both active [Riehle. 1985]. Riehle [1985] found 35 tephra layers for eruptions from Crater Peak over the last 6.000 years, which provides a minimum indication of the long term eruption frequency of Crater Peak. The Crater Peak vent has produced two eruptions this century: a single eruptive event that occurred on July 9. 1953 [Juhle and Coulter. 1955], and the sequence of three eruptions that occurred during 1992 [AVO Staff, 1993]. The total erupted volume for the 1992 Crater Peak sequence is approximately 41 x 106 m ' DRE {dense rock equivalent) of magma [Neal et al.. 1995]. The magma erupted in 1992 is a calcalkaiine andesite. containing about 57 wt.% SiO: by weight [Nye et al., 1995]. Magma of this composition is typical for Crater Peak eruptions, which are generally more

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 3

mafic than the neighboring volcanoes in the Cook Inlet region (Figure 3.2). Deposits from all three 1992 eruptions are composed of both a tan and gray colored andesite that are found to differ only in the composition of the groundmass glass [Harbin et al., 1995]. Gardner et al., [in press] also describe a tan or brown and a gray colored andesite present in tephra samples that differ in the size and volume percentage of mirolites. Based on the similarities in composition, all three 1992 eruptions are thought to have tapped the same magma reservoir. Several lines of evidence suggest that the 1992 Crater Peak andesitic magma differentiated from its parent at lower crustal depths. A section of the olivine-quartz- plagioclase-diopside tetrahedron with samples from the 1992 eruption of Crater Peak, the 1989-90 eruption of Redoubt Volcano, and representative samples from recent eruptions of Augustine Volcano for comparison is shown in Figure 3.3. These data suggest that Crater Peak magmas equilibrated at higher pressure and therefore greater depth than other Cook Inlet magmas. Harbin et al.. [1995] found pristine hornblende crystals in the 1992 magmas, suggesting it did not spend significant time shallow enough to be outside the hornblende stability field. Comparison between the 1953 and 1992 magmas reveals a number of incompatible trace element differences [Nrye et al.. 1995], suggesting they were derived from separate sources. Gardner et al.. [in press] calculated magma discharge rates, magma supply rates, and magma ascent velocities using calculated eruptive volumes, eruption durations, ranges of estimated magma discharge rates, as well as estimated depths of the magma source for each of the three eruptions. Estimated magma discharge rates ranged from 823 to 1157 m 'Y 1. while magma supply rates were 145 to 867 mJ/s. Magma ascent velocities ranged from 0.06 to 0.9 m/s. with a mean of 0.4 m/s. Using the calculated values for magma supply rate and ascent velocity. Gardner et al.. [in press] estimated a conduit radius of about 21 m using the methods described by Scandone and Malone [1985]. Prior to the 1953 eruption, the Crater Peak vent was filled with ice. Following the 1953 eruptions, a shallow lake formed within the Crater Peak vent. The lake water appeared blue-green in color, and was sampled during the summer of 1970 [Keith et al.. 1995]. Visual observations suggest that the crater lake remained largely unaffected through most of the precursory seismic activity until June 8. 1992. when it turned a dark gray color with several visible areas of strong upwelling. The lake water was sampled on June 11. and geysering was

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44

observed in areas surrounding the lake for the first time. The 1992 water samples were found to be higher in S04 than those collected in 1970. This change in chemistry was attributed to increased sulfur species released from rising magma [Keith et al.. 1995], By June 26, 1992, the water of the crater lake had completely disappeared leaving only a gray mud and substantial fumarolic activity. Gas emissions from the Crater Peak vent were measured by a number of techniques during the 1992 eruption sequence. These included Total Ozone Mapping Spectometer (TOMS), Airborne Ultraviolet Correlation Spectrometer (COSPEC), Airborne Infrared Spectrometer (MIRAN). visual observations, and direct sampling of the lake water in Crater Peak. Large plumes containing 230 x 70. 400 ± 120. and 200 ± 60 kilotons of SO: were detected by the TOMS system for the June. August, and September eruptions respectively [Bluth et al.. 1995]. While the individual eruptions produced large amounts of SO:, measurements using the COSPEC found only minor SO: flux during the precursory and inter- eruptive periods. The SO: fluxes determined using COSPEC data ranged from ten to hundreds of tons per day between September 21 and October 10, 1992; CO: fluxes determined by the MIRAN measurements yielded values of thousands to tens of thousands of tons per day between September 25. 1992 and January 21, 1993 [Doukas and Gerlach. 1995]. Doukas and Gerlach [1995], suggest that the low flux of SO: observed through the precursory and eruptive sequence, likely results from the scrubbing of SO: by water as magmatic gases passed through the Crater Peak hydrothermal system. Higher SO: values observed after the September 16-17 eruption were attributed to drying or disruption of the hydrothermal system during the eruption. The hydrolysis of SO: in the Crater Peak hydrothermal system explained the H2S odor observed during periods when the COSPEC detected only background amounts of SO:, the observed decline of SO: during periods of tremor in October 1992. and the observed increase in sulfates in the Crater Lake prior to the June 27 eruption. During the three eruptions, large fluxes of SO: were attributed to large magma flux that overpowered the ability of the hydrothermal system to scrub SO: [Doukas and Gerlach. 1995]. Alternatively, Gardner et al.. [in press] suggest that the low S02 flux may result from the absence of magma at shallow levels as well as from the presence of a plug at shallow depth.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45

3.3.2 Seismicity and Geophysics Earthquake hypocenters between 1981 and 1992 were observed to form four main clusters: beneath the north caldera rim, under the summit of Mount Spurr. beneath Crater Peak, and south of the Chakachatna River [Jolly et al.. 1994], The activity beneath the summit was roughly constant with time, whereas there were individual swarms under the north caldera rim in 1982 and 1989. The area beneath Crater Peak was aseismic until precursory activity began in August 1991 (Figure 3.4). Seismic activity south of the Chakachatna River was roughly constant with time and is probably not directly associated with volcanic processes at Mount Spurr. Jolly et al.. [1994] also identified a regional shoaling of earthquake hypocenters beneath Mount Spurr. In areas surrounding the volcano, shallow crastal seismicity was observed at depths as great as 15-20 km, whereas the area at 5 - 20 km depth directly beneath the volcano was aseismic. This aseismic zone was attributed to a warm ductile area likely resulting from repeated intrusive activity beneath the volcano. Power et al., [1995] give a preliminary seismic chronology and a review of the use of seismic observations in formulating eruption forecasts. A histogram showing the number of located seismic events per day and the various individual swarms is shown in Figure 3.4. The space-time development of earthquake hypocenters during 1991 and 1992 suggests that the precursory seismic activity resulted from the intrusion of magma at depths of 5 to 15 km. The aiignment of deeper hypocenters at 20 - 40 km depth was attributed to magma transport and stress adjustments. The November 9-10 and December swarms were interpreted as intrusions of magma that did not reach the surface [Power et al., 1995]. Wiemer and McNutt [1997] calculated the spatial distribution of 6-values beneath Crater Peak. They identified a prominent zone of high 6-values (6-1.7) between 2.3 and 4.5 km depth beneath Crater Peak. They also identified a less prominent increase in 6 (6-1.2) below about 10 km depth. Using hypocenters from 80 VT earthquakes between 20 and 40 km depth, they obtained a 6-value of 1.67. They attributed the 6-value anomaly between 2.3 and 4.5 km depth to the fracturing of rock caused by vesiculation of magma, and suggested that the higher 6-values below 10 km resulted from the possible presence of a magma reservoir. The higher 6-values at depths of 20 - 40 km were attributed to a deep magma chamber.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46

Volcanic tremor was first observed on June 6. 1992. as short “bursts'’ that lasted several minutes [McNutt et al.. 19951- Isolated instances of tremor increased in strength and duration through June. Two strong episodes of tremor occurred on June 24 and June 25 and lasted 154 and 142 minutes respectively. Tremor occurred continuously for 23 hours prior to the June 27 eruption. Calculated tremor amplitude ratios suggest that the source of tremor migrated to shallower depths between June 11 and June 24 [McNutt et al.. 1995]. Short episodes of tremor lasting 12 and 11 minutes, respectively occurred prior to the August 18 and September 16-17 eruptions. Tremor of variable amplitude was also recorded between the September 16-17 eruption and October 20. Episodes of post-eruptive tremor were particularly strong between October 1 and 6. and were attributed to degassing of magma at shallow depths [McNutt et al.. 1995: Gardner et al.. in press]. Using P-wave arrival times for locally recorded earthquakes. Power et al., [in press] determined a three-dimensional P-wave velocity structure to depths of 10 km beneath Mount Spurr. Results from the study show a prominent Iow-velocity zone extending from the surface to 3-4 km depth beneath the south flank of Crater Peak, a narrow low-velocity zone extending to at least 10 km depth to the southeast of Crater Peak, and a less pronounced low- velocity zone beneath the summit of Mount Spurr. These low-velocity areas were attributed to a shallow hydrothermal system, a magmatic conduit extending from 10 km depth to the Crater Peak vent, and a fractured or warm area beneath the summit of Mount Spurr. respectively. This study found no large low-velocity zones indicative of a magma chamber within the upper 10 km of the crust beneath Mount Spurr. Turner and Wescott [1986] and Wescott et al.. [1988] had previously identified the hydrothermal system to the south and southeast of Crater Peak using a number of geophysical exploration techniques.

3.3 Data and Earthquake Classification

3.3.1 Seismic Instrumentation and Recording Between January' I. 199! and December 31. 1993. Mount Spurr was monitored by a network of 6 to 9 short-period radio-telemetered seismic stations (Figure 3.5). Data from this network were collected using an event-detected recording system that digitizes incoming signals at 100 sampies per second [Lee et al.. 1989]. Details of the network operation and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47

acquisition system are given by Jolly et ai., [1996], Using this system. P- and S-wave picking errors are roughly ±0.01 s and -0.05 s respectively. Hypocenters and local magnitudes, determined by frequency and amplitude measurements, were calculated using the program HYPOELLIPSE [Lahr. 1989]. In this study, we use the velocity model and station corrections developed for the Mount Spurr area by Jolly et al.. [1994]. During the period of this study, about 2.500 earthquakes were located within 20 km of the summit of Mount Spurr. The estimated precision (or relative location uncertainty between neighboring events) of earthquake hypocenters at Mount Spurr is - 1 km in epicenter and 2 km in depth for events at depths of less than 10 km. The precision is about 3 km in epicenter and 2.5 km in depth for events at depths greater than 20 km. These estimates are based on average values for subsets of located events using the standard errors calculated by HYPOELLIPSE [Lahr. 1989].

3.3.2 Classification Method VT earthquakes. LP events, and hybrid events are distinguished by the frequency content of the observed waveform, examples of each are shown in Figure 3.6. In classifying seismic events in this manner, it is necessary to determine whether observed spectra are a result of the source or a secondary wave propagation effect, such as attenuation, scattering, or diffraction between the source and seismic station. To eliminate path effects, we have calculated velocity spectra for each of the representative event types at five stations at varying distances and azimuths from the hypocenter. The spectra were generated using a 20.48-second window beginning approximately 2 seconds before the onset of the first P-arrival. Each of these individual spectra were then normalized to one. stacked together, and then normalized a second time. By stacking spectra in this manner, any unusual path effects between a source and receiver will be eliminated in the stacking process. Pitt and Hill [1994] used a similar technique to differentiate between VT and LP events at Long Valley Caldera. California. The stacked spectra from VT earthquakes at Mount Spurr generally show substantial energy from approximately 2 Hz to between 6 and 8 Hz. In contrast. LP events normally show a strong peak between I and 3 Hz. with little energy above 3.5 Hz. The stacked spectra for hybrid events generally show a strong pulse close to 2 Hz and a slow decline in energy to 5 or 6 Hz. Examples of the stacked spectra of a VT earthquake at 35 km depth, an LP event at 38 km depth, and a hybrid event at 22 km depth are shown in Figure 3.7.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48

Once the spectral character and waveform appearance of individual event types were determined, we classified all located events as VT. LP. hybrid, or unknown using the following procedure. First, waveforms from individual events were classified by visual inspection. This was usually the case for shallow events as path effects are typically small where the hypocenters are close to the recording stations. For events with hypocentral depths greater than 10 to 15 km. stacked spectra were generally required, as path effects are more prevalent with the longer travel paths. Events that lacked a reliable spectrum as a result of telemetry noise, magnitude, or poor station geometry were classified as unknown.

3.4 Results and Interpretation

The results of the event classification are summarized in an epicenter map and a cross section that show the three event types (events classified as unknown are not plotted) (Figure 3.8). and a plot of focal depth versus time (Figure 3.9). A summary of magnitudes of the various event types and the cumulative seismic energy release is shown in Figure 3.10. Both the classification of seismic events and existing geophysical data are used to develop a simplified model of the Crater Peak magmatic system and the volcanic processes that are likely to have been active through the 1992 eruption sequence.

3.4.1 St ay matte System Geometry We speculate that the magmatic system beneath Crater Peak is composed of a source zone whose seismic expression extends from 20 - 40 km depth, a small reservoir or storage area whose top is at about 10 km depth, and a pipe-like conduit system extending from the reservoir to the surface. The geometry of these bodies is imaged by event locations and for shallower features, by the three-dimensional P-wave velocity structure determined by tomographic inversion of local earthquake data [Power et al.. in press]. The deep magma source zone at 20 - 40 km is best defined by the mix of VTs. LPs. and hybrids located to the southeast of Crater Peak (Figure 3.8), which peaked after eruptive activity ended. This is consistent with petrologic evidence (Figure 3.3) suggesting that the 1992 magmas were derived deep in the crust or uppermost mantle [Nye et al.. 1995]. The mid-crustal magma reservoir or storage area is best defined by the energetic swarm of VT

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49

earthquakes that occurred at the close of the September 16-17 eruption at 5 to 11 km depth (Figure 3.1 ID). Unfortunately, only 23 earthquakes of the hundreds that occurred in this swarm can be reliably located as the ongoing eruption tremor masked signals on the local stations. Storage of magma at mid-crustal depths is consistent with pristine hornblende crystals found in the material erupted in 1992 [Harbin et al.. 1995]. Based on the small volume of erupted material in 1992. we presume the magma is stored in a small or rather diffuse volume. This small mid-crustal reservoir may be nothing more than a widening of the conduit system, perhaps a network of interconnected dikes and sills. The conduit between 10 km depth and the surface is imaged by hypocenters (Figure 3.8) and seismic tomography, which shows a narrow [ow-velocitv zone extending to at least 10 km depth to the southeast of Crater Peak [Power et al.. in press]. Significant swarms of VT earthquakes, which locate in the upper conduit and around the mid-crustal storage area, occur during August 1991. June 5. June 27. November 9-10. and December 23 - 29. 1992 (Figure 3.11A. B. C. E). While much of the precursory sequence occurs in this depth range (Figure 3.9). these individual swarms occur when related evidence suggests that magma or volatiles might have been moving to shallower depth. Particularly the June 5 and June 27. 1992. swarms, that occurred the day prior to the first shallow volcanic tremor [McNutt et al.. 1995] and immediately preceded the first eruption. An idealized sketch of the inferred components of the Mount Spurr/Crater Peak magmatic plumbing system is shown in Figure 3.12. Prior to the June 1992 eruption, hypocenters were observed beneath the summit of Mount Spurr and under the north caidera rim. A low-velocity zone imaged by seismic tomography is suggestive of a fractured or altered area beneath the summit of Mount Spurr. perhaps a remnant magma chamber or conduit system which fed this vent. The earthquake hypocenters beneath the north caidera rim may represent some type of structural control in this area, such as a fault system, although high elevation and extensive glaciation limit seismic and geologic observations in this area. We assume that both these areas were responding to perturbations in the local stress field resulting from the emplacement of magma beneath Crater Peak. Evidence for the hydrothermal system on the southeast flank of Crater Peak includes results from resistivity, self potential, and mercury and helium soil analyses [Turner and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50

Wescott. 1986], seismic tomography [Power et al.. in press], and the presence of hot springs and active fumaroies at Crater Peak. The seismic expression of the hydrothermal system during the 1992 eruption sequence is apparently limited to volcanic tremor [McNutt et al.. 1995]. This hydrothermal system may contribute to the low 6-value anomaly between 2.3 and 4.5 km depth identified by Wiemer and McNutt [1997], Each of the eruptions lasted 3.5 to 4.0 hours [McNutt et al., 1995] and produced roughly equivalent amounts of erupted material [Neal et al.. 1995], Geochemical and petrologic studies suggest that all three eruptions tapped the same reservoir [Nye et al.. 1995. Harbin et al.. 1995: Neal et al.. 1995: Gardner et al.. in press], that we think is at about 10 km depth based on seismicity i Figure 3.11C and 3.11 FT Interpretations presented here suggest little change to the value calculated by Gardner et al.. [in press] for magma ascent rate, except preference might be given to values which used a depth of 10 km for the magmatic source.

3.4.2 Seismic Activation and Evolution Almost all of the events located during the precursory seismic sequence were VT earthquakes. This includes the swarm beneath Crater Peak in August 1991. the caldera-wide increase in activity between November 1991 and June 1992. as well as the earthquakes in the 20-to-40-km depth ranee prior to June 1992 (Figure 3.9). The August 1991 swarm was the first significant number of earthquake hypocenters observed beneath Crater Peak in more than 10 years of monitoring [Jolly et al.. 1994], We speculate this seismicity represents a stress response as magma from a deeper source moved to the mid-crustal storage area. The caldera- wide increase in seismicity between November 1991 and June 1992 likely resulted from continued pressunzation of this mid-crustal magma storage area. The only seismic expression of the deeper source zone during the precursory period are the 18 VT earthquakes observed at 20-to-30-km depth and the single hybrid event at 32 km depth on June 18. 1992 (Figure 3.9). Lack of increased SO; flux [Doukas and Gerlach. 1995], and no marked increase in fumorolic activity or observed changes in the crater lake support the assumption that any magma in the system remained at some depth between August 1991 and early June 1992. Several independent lines of evidence suggest that magmatic volatiles. perhaps accompanied by a small amount of magma, began to ascend from the mid-crustal storage zone in early June. On June 5. the largest number of located VT earthquakes in any 24-hour period

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51

occurred beneath Crater Peak (Figure 3.1 IB). Following June 5. the number of locatable earthquakes declined throughout the caidera until immediately prior to the June 27 eruption. We attribute this decline in seismicity to a concentration of stresses around ascending magma or volatiles as they moved to shallower depth beneath Crater Peak. On June 6. shallow volcanic tremor, attributed to the interaction of ground water and magma or related volatiles. was observed for the first time. The occurrence and intensity of tremor increased through the month of June [McNutt et al.. 1995]. Additionally, beginning in mid-June, visual observations of increased fumorolic activity, changes in the crater lake water color and chemistry, and upwelling in the lake are all suggestive of increased heat flow possibly associated with the ascent of magma and associated volatiles. Shallow LP events were not located until June 19. A single hybrid event was identified at 32 km depth on June 18. 1992. 9 days before the June 27 eruption. The June 27 eruption was preceded by approximately 19 hours of continuous strong volcanic tremor [McNutt et al.. 1995] and a 4 hour long swarm of VT earthquakes that preceded the onset of eruptive activity. These shocks were generally located between -I and 4 km depth directly beneath the Crater Peak vent (Figure 3.11C). Magnitudes of these events ranged from -0.6 to 1.2 and generally increased in size throughout the duration of the swarm. Within this swarm, four events were determined to be LPs. Presumably, this seismic activity represents the final opening of the conduit and ascent of magma to the surface. Following the June 27 eruption, shallow seismicity quickly declined. A few LP events at depths above sea level were recorded in the 48-hour period following the eruption, but these quickly died away. At shallow depths, the period between the June 27 and September 16-17 eruption was as quiet seismically as any period since the onset of precursory seismicity m August of 1991. Activity between 10 and 40 km depth increased slightly between the June 27 and August 18 eruption. An LP event, a hybrid event, and 5 VT earthquakes were identified during this period (Figure 3.9). The August 18 eruption initiated a further increase in all event types in the 20 - 40 km depth range. This activity continued to increase until about 2 weeks after the September 16-17 eruption. Interestingly, little change was observed in the shallow seismic activity either before or after the August 18 eruption (Figure 3.9). The August and September eruptions were preceded by much subtler immediate precursors: presumably, the upper portions of the conduit were now open allowing the unimpeded ascent

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52

of magma. The period between each of the eruptions was quiet seismicaily at shallow depths (Figure 3.9), suggesting little movement of magma between eruptions above 10 km depth. We think the strong swarm of VT earthquakes that began during and followed the September 16-17 eruption, reflects a stress adjustment resulting from the removal of magma at about 10 km depth (Figure 3.1 ID). This was the most energetic swarm of seismic events associated with the 1992 eruption sequence (Figure 3.10). Presumably, the removal of mass by the three eruptions combined was sufficient to initiate this type of stress response, while the mass removed by the individual eruptions in June and August was not.

3.4.3 Deep Seismicity and Eruption Termination One of the most unusual aspects of seismicity observed at Mount Spurr in association with the 1992 eruption sequence is the initiation of activity between 20 and 40 km depth. VT earthquakes below 20 km depth began to appear as early as August 1991, roughly coincident with the onset of shallow VT earthquakes beneath Crater Peak. This activity intensified slightly in January 1992. although the total number of VT earthquakes located in this depth range during the precursory period is small (Figure 3.9). The June eruptive event triggered an increase in activity below 20 km depth: the first locatable deep LP event occurred on July 27 at a depth of 35 km. The rate of deep LP seismicity increased again following the August 18 eruption and reached a peak during late October (Figure 3.9). LP events at depths greater than 10 km have been observed at only a few volcanoes. These include Kilauea Volcano. Hawaii [Koyanagi et al.. 1987], Long Valley Caldera. California [Pitt and Hill. 1994], Mount Pinatubo. Philippines [White. 1996], Mount Lassen. California [Walter. 1991]. Mount Morivoshi. Japan [Hasagawa et al.. 1991]. Mount Fuji. Japan [Ukawa. 1997], and Izu-Ooshima. Japan [Ukawa and Ohtake. 1987], A handful of isolated events have also been reported at several other volcanoes in the Cascades [Pitt and Hill. 1994: Malone and Moran. 1997], White [1996] gives a review of observations of deep LP events and a discussion of possible source mechanisms. In many of these occurrences of deep LP events, the fluid most likely responsible for their generation is basaltic magma, which is thought to be vapor saturated with CO: [White. 1996]. Unlike occurrences of deep LP activity at Mount Pinatubo and Long Valley, events at 20 - 40 km depth at Mount Spurr are a mix of VT. LP. and hybrid events. The mix of seismic

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53

event types seen at Mount Spurr between 20 and 40 km depth is most analogous to that seen at Kilauea at similar depths. White and Dzurisin [1997] suggest that deep LP activity may be triggered when the shallow portions of a magmatic system are depressurized either by an eruption, differentiation of magma in a shallow magma chamber, or stress changes associated with large earthquakes. That we do not see seismic activity in the 20 - 40 km depth range at neighboring active Cook Inlet volcanoes, such as Redoubt and Augustine (Figure 3.1). is perhaps a result of the shallow or mid-crustal magma chambers which are present at these volcanoes [Figure 3.3. Power et al.. 1994; Power 1988], which apparently does not exist at Crater Peak. We infer that the deep seismicity observed at Mount Spurr is likely the result of the movement of magma at depths of 20 - 40 km. This activity apparently started with VT earthquakes that began about the same time as the August 1991 Crater Peak swarm, that we attribute to the initial rise of magma and pressurization of the mid-crustal storage zone. The observed increase in seismicity in the 20 - 40 km depth range may represent an increase in the movement of magma from the deeper zone, as magma was removed from the mid-crustal storage zone by each of the three eruptions. Assuming that the seismicity rate in the 20 - 40 km depth range reflects the rate of magma transport, then maximum flux occurred in late October or early November. 1992. We believe the November 9-10. 1992 earthquake swarm most likely reflects a magmatic intrusion, as suggested by Power et al.. [1995]. Most shocks in the November 9-10 swarm came from one of three individual earthquake families (C. Stephens, pers. comm. 1993]. Each of these families was activated as magma presumably moved progressively farther from the existing conduit. We speculate that the magma intruded on November 9-10. may have come from greater depth, contained fewer volatiles. and had a higher viscosity than the magma erupted during the summer. This magma was unable to reach the surface and instead intruded at shallow depths causing the energetic earthquake swarm. Supporting circumstantial evidence comes from the peak rates of seismic activity observed in 20 - 40 km depth range prior to November 9-10. The magma emplaced on November 9-10 may have blocked the conduit for the rise of additional magma and ended the 1992 eruption sequence. Alternatively. Gardner et al.. [in press] suggest that the 1992 eruption sequence was terminated when additional magma rose, crystallized, and blocked the conduit following the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54

September 16-17 eruption. The strong volcanic tremor in September and October 1992 [McNutt et al.. 1995] and the relatively high gas flux [Doukas and Gerlach. 1995] are suggestive of the emplacement of shallow magma or a disruption of the hydrothermal system following the September 16-17 eruption. Based on the cyclic nature of the tremor in September and October. Benoit [1998] suggests this tremor is the result of a hydrothemal process. We favor eruption termination caused by influx of new magma on November 9 — 10. A later, less energetic swarm of VT earthquakes occurred between December 23 and 29. 1992. Hypocenters in this swarm are clustered between 7 and 10 km depth, in an area roughly coincident with the September 17 swarm (Figures 3.1 ID and 3.1 IF). Possibly, this seismicity reflects a small intrusion or stress adjustment associated with the mid-crustal magma storage zone. Throughout 1993. seismic activity at shallow depth remained almost exclusively VT earthquakes concentrated between sea level and 5 km depth. Deeper activity slowly declined through 1993. and deep LP and hybrid events declined more rapidly than VT earthquakes (Figure 3.9). This slow decline in seismic activity is often observed at active volcanoes following the cessation of eruptive activity [Mori et al.. 1996).

3.4.4 Eruption Forecasting and Volcanic Hazards Careful monitoring of seismicity, as well as visual observations of the volcano and measurements of gas emissions, allowed the Alaska Volcano Observatory t.AVO) to issue forecasts of impending eruptive activity prior to the June 27 and September 16-17 eruption, and notifications were issued shortly after all three 1992 eruptions began. Two forecasts were issued when eruptions did not occur, based on increased tremor in October 1992 and during the November 9-10 swarm. A synopsis of AVO’s forecasting efforts and public statements is given by Eichelberger et al.. [1995]. Based on the geologic record and the lack of a shallow magma chamber that could feed larger eruptions, we expect future eruptions of Crater Peak to be vulcanian to subplininan events similar to those in 1953 and 1992, with precursory seismicity perhaps similar to that seen m 1991 and 1992. The shallow seismic precursors to the August 18 and September 16-17 eruption were much shorter than those on June 27. Consequently, close monitoring should follow any future eruptive event of Crater Peak for at least several months.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55

It should also be remembered that energetic swarms of seismic events such as November 9-10 and December 23-29. and energetic periods of tremor did not result in eruptions. The summit vent of Mount Spurr has not erupted for 5.000 to 6,000 years [Riehle. 1985: Nye and Turner, 1990]. Reactivation of that vent would be an unusual event and we might see a very different pattern of seismicity than in 1992.

3.5 Conclusions The occurrence of VT earthquakes. LP events, and hybrid events indicates that a variety of seismic sources were activated by the 1992 eruption sequence of Crater Peak. Stacking of normalized velocity spectra suggests the observed differences in waveforms reflect source processes, and are not the result of wave propagation effects such as attenuation, scattering, or diffraction. Both seismic and geochemical evidence suggests that magma involved in the 1992 eruptions originated at 20 - 40 km depth. This magma then migrated to a small mid-crustal storage zone that was tapped for all three eruptive events in 1992. Earthquake hypocenters suggest the conduit dips to the southeast of Crater Peak. The long duration widespread precursory seismicity likely reflects perturbations in the local stress field associated with the movement of this magma into the mid-crustal storage zone at about 10 km depth. We think that an intrusion of more viscous magma occurred on November 9-10 and blocked the upper portions of the conduit ending the 1992 eruption sequence.

3.6 Acknowledgments

This manuscript benefited greatly from discussions with Cynthia Gardner. Randy White. Tina Neal. Chris Stephens, and Seth Moran. Harley Benz and Cynthia Gardner provided helpful reviews of the text and figures. Dave Sentman made many helpful suggestions for calculating the stacked velocity spectra.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56

148°W0 156 W 61 N 61 N Mount Hayes Volcano Mount Spurr 4 Crater Peak

100 km | o 59 N o 59 N o 156 W 148° W

Figure 3.1) Location Map for Mount Spurr: Map showing the Cook Inlet region of Alaska. Mount Spurr and Crater Peak, neighboring volcanoes (solid triangles), and nearby communities.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57

Range of SiO 2 in Cook Iniet Volcanoes

70 - 0) 8 60 - “ Augustine >» —(0 50 - □ Redoubt ^ 40 - ■ Crater Peak o 3 0 g> 20 - JQ E 1 0 3 v/, z 0 - 50 52 54 56 58 60 62 64 66 68 72

Wt. % SiO 2

Figure 3.2) Range of silica observed at Cook Inlet volcanoes: Range of silica observed in erupted products at Crater Peak. Redoubt Volcano, and Augustine Volcano.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58

50% Diopside

Figure 3.3) Olivine - Quartz - Plagioclase- Diopside tetrahedron for Cook Inlet: A section of the olivine-quartz-plagioclase-diopside tetrahedron showing the projection of samples from Crater Peak (circles). Augustine Volcano (squares), and the 1989-90 eruption of Redoubt Volcano (triangles). Solid lines represent position of the 1 atmosphere cotectics and stars show the location of the 10 and 25 Kbar eutectic. The Crater Peak magmas have equilibrated at higher pressure and presumably greater depth than neighboring volcanoes.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59

150 L_L_L_1_! I I I— I I— I— I— 1— L. 140 130 > 120 < a 110 x r-rl. 100 90 z 80 > a 70 o 60 X 50 3 40 2 30 z 20 10 0

j j v ? a N A \I J 'j 'A S O N* O t 1991 1992 1993

Figure 3.4) Histogram of located earthquakes at Mount Spurr. 1991 - 1993: Histogram showing the number of earthquakes located within approximately 18 km of the summit of Mount Spurr between January 1. 1991 and December 31. 1993. Arrows correspond to eruptions on June 27. August 18. and September 16-17. 1992.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60

Figure 3.5) Map of seismic stations at Mount Spurr: Seismic stations (solid triangles) located in the vicinity of Mount Spurr. Solid line denotes the outline of the rim of the Mount Spurr caidera.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61

500 V T 06/5/92 Z = 1 km

o ------

1000 LP t i M ' . , 6/28/92 Z = -1 km o /.v/v; 's'A . v\,

iooo VT 11 12/23/92, Z = 38 km r* *»•! k Z o 3 O -1 0 0 0 u

LP > ( 12/22/92 Z = 35 km 1 0 0 0 ... i * - i y i. >.. '. a vi » . . , > _ L l> 1 . ..^..1 - ■ ■ ■■ * ‘A l: , 4 / * .' ■•••, j. - - 0 , r • - . • * ■; < tt +. • • • ', -./* ' *: «*»■. >.-/■- - y , * -i ' \ ■ • ' ’ ,r ‘ ' ■ .• ' ' " -1 0 0 0

iooo HYBRID k , i l . i 10/17/92 Z = 22 km o .".A ■„ <* ' -1 0 0 0 K V - « V -' 0 10 15 20 25 30 SECONDS

Figure 3.6) Waveforms of representative seismic events: Waveforms from representative seismic events located at various depths, as recorded at station CGL.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62

as H 5J Ed eu

>« r j U O a > Q Ed N •>» < s a* W z

Figure 3.7) Normalized velocity spectra of LP. VT. and hybrid events. Spectra are calculated at 5 individual stations in the Spurr network and stacked together to eliminate path effects. Waveforms lor the individual events are shown in Figure 3.6.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63

A’

VS • ~ 20 l

rvi 3 30 - o _ ■» V ^ b» * o * ■ « ‘ J* *

0 10 DISTANCE (KM)

Figure 3.8) Epicenter map and cross section of located earthquakes near Mount Spurr: Epicenter map and cross section showing all located earthquakes near Mount Spurr from 1991 through 1993. Only epicenters within the box are shown in cross section A-A'.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64

J FMAMJ JASONDJ FMAMJJASONDJFMAMJJASOND 1991 1992 1993

Figure 3.9) Focal depth versus time for all located events at Mount Spurr. 1991 - 1993. Arrows and dotted lines note the approximate times of the June 27. August 18. and September 16-17 eruption. The offset in hypocentral depths from late February to mid-March. 1992 is a result of telemetry problems with several stations in the Spurr network.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65

J I I— L _l I ! I L _ l I 1 I L 2 1 a 0 a 2 LP MAGNITUDES Z 1

0 2 HYBRID MAGNITUDES

1 2 0 I I ll I i l 1 ll i !>■ I i II "i M-t CUMULATIVE SEISMIC November 9 - 10 CZ2 ENERGY RELEASE Swarm O aeS 10 H Precursory September 16-17 Buildup Swarm August 1991 Swarm D 0 1— i— :— i— :— i— i— :— i— i— i— :— I— i— i— i— i— i— I— i— i— i— :— :— I------:— 1— i— ;— i— i— i— i— r i r J FMAM J JASONDJ FM AM JJASONDJ FMAM J J A S O N D 1991 1992 1993

Figure 3.10) Magnitudes and cumulative seismic energy release at Mount Spurr. 1991 - 1993: Magnitudes of A) located VT earthquakes. B) LP events. C) hybrid events, and D) cumulative seismic energy release between 1991 and 1993. Note the large relative increase in seismic energy release associated with the VT swarm at the close of the September 16-17 eruption. Arrows correspond to the approximate times of the June 27. August 18. and September 16-17 eruption.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ■t —yr — —'* - ■ ...... II -.i . « • f ' ■ m ' i : 3 . ; i ; ; t ’ ; . ;>S* - . . ~

i -

8/91 6/5/92 6/27/92 9/17/92 11/9/92 - 12/23/92 - 11/10/92 12/29/92

MAGNITUDE SCALE c -9.0+- Z 0 0 Z *o+-

2. 0 +-

Figure 3.II) Epicenter maps and cross-sections for shallow swarms at Crater Peak: VT earthquake epicenter maps and vertical cross sections for profile B - B’ for: A) August 1991. B) June 5. 1992. C) June 27. 1992. D) September 17. 1992. E) November 9 - 10. 1992. F) December 23 - 29. 1992. For comparative purposes earthquake hypocenters are located with a consistent set of stations that did not include either stations CPK or CP2 (Figure 3.5). Plots A. B. C. and E define the upper conduit, while plots D and F show the probable location of the mid-crustal storage zone. Plot D shows only the 23 best located earthquakes of the hundreds that occurred on September 17. 1992.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67

VT earthquakes beneath summit of Mount Spurr and N. caldera rim. .reflecting increased stress caused by intrusion of magma to mid-crustal Approximate extent of Crater storage zone. Peak hydrothemal system (Turner and Wescott, 1985. Upper magma conduit as defined Power etai. 1998). by swarms of VT earthquakes in 8/91.6/5. and 6/27,1992 I Figure 11) and seismic tomography i Power et al. 1998). Conduit diameter estimated to be 21 m or less (Gardner et al. 1998).

Mid-cnislai level magma storage * zone as defined by 9/17 and 12/24 - 29. hypocenters (Figures 11C and 1 IF) and pristine hornblende crystals (Harbin et al. 1995).

Deep crustal magma source area as defined by mix of VT. LP. and Hybrid events. Chemical phase data also suggest 1992 magma equilibrated in the lower crust or upper most mantle (Figure J and Nye et al. 1995).

DISTANCE (KM)

Figure 3.12) Sketch of the inferred Crater Peak magmatic plumbing system: Inferred Crater Peak magmatic system in NW to SE cross section. Symbols (X. O. and A) refer to approximate locations of located VT. LP. and hybrid events. Question marks indicate uncertainty in extent of Crater Peak hydrothermal system and the shape and extent of the mid- crustal storage zone. Location of cross section B-B* is shown in Figure 3.11.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68

3.7 References

Aki, K. and R.Y. Koyanagi, Deep volcanic tremor and magma ascent mechanism under Kilauea. Hawaii. J. Geophys. Res., 89. 7095-7109. 1981.

Alaska Volcano Observatory Staff. Mount Spurr's 1992 eruptions. EOS Trans., 74, 217-222. 1993.

Benoit. J.P.. Pattern and process in volcano seismology. Ph.D. dissertation. University of Alaska - Fairbanks. 155 pp.. 1998.

Bluth. G.J.S.. C.J. Scott. I.E. Sprod. C.C. Schnetzler. A. J. Krueger, and L. S. Walter. Explosive emissions of sulfur dioxide from the 1992 Crater Peak eruptions. Mount Spurr Volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano, Alaska, edited by T.E.C. Keith. USGS Bull. 2139. 37-46. 1995.

Chouet. B.A.. A seismic model for the source of long-period events and harmonic tremor, in Volcani Seismology’. LAVCEI Proc. in Volcanology, edited by P. Gasparini. R. Scarpa, and K. Aki. Springer Verlag. Berlin. 3. pp 133-156. 1992.

Chouet. B. A.. New methods and future trends in seismological volcano monitoring, in Monitoring and mitigation of volcano hazards, edited by R. Scarpa and R.I. Tilling, Springer. Berlin, pp 23-98. 1996.

Chouet, B.A.. R.A. Page. C.D. Stephens. J.C. Lahr, and J.A. Power, Precursory’ swarms of long-period events at Redoubt Volcano (1989-90). Alaska: their origin and use as a forecasting tool. J. Vole, and Geotherm. Res.. 62. 95-136. 1994.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69

Doukas, M.P.. and T. Gerlach. Sulfur dioxide scrubbing during the 1992 eruptions of Crater Peak. Mount Spurr volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bui’.. 2139. 47-58. 1995.

Eichelberger. J.E.. T.E.C. Keith. T.P. Miller, and C J. Nye. The 1992 eruptions of Crater Peak Vent. Mount Spurr volcano. Alaska: chronology and summary, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bull. 2139. 1 18, 1995.

Gardner. C.A.. K.V. Cashman. and C.A. Neal. Tephra-fall deposits from the 1992 eruption of Crater Peak. Alaska: implications of clast textures for eruptive processes. Bull, of Volcanol. In press. 1998.

Harbin. M.L.. S.E. Swanson. CJ. Nye, and T.P. Miller. Preliminary petrology and chemistry of proximal eruptive products: 1992 eruptions of Crater Peak. Mount Spurr Volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T. E. C. Keith. USGS Bull. 2139. 139-148. 1995.

Harlow. D.H.. J.A. Power. E.P. Laguerta. G. Ambubuyog. R.A. White, and R.P. Hoblitt. Precursory seismicity and forecasting of the June 15. 1991. eruption of Mount Pinatubo. in Fire and Mud. eruptions and lahars of Mount Pinatubo. Philippines, edited by C. G. Newhall and R. S. Punongbuvan. University of Washington Press. Seattle. WA. p. 285-306. 1996.

Hasagawa. A.. D. Zhao. S. Hori. A. Yamamoto, and S. Horiuchi. Deep structure of the northeast Japan arc and its relationship to seismic and volcanic activity. Nature. 352. 683-689. 1991.

Jolly. A.D.. R.A. Page, and J.A. Power. Seismicity and stress in the vicinity of Mount Spurr volcano, south, central Alaska. J. Geophys. Res.. 99. 15305-15318. 1994.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70

Jolly. A.D.. J.A. Power. S.D. Stihler. L.N. Rao. G. Davidson. J.F. Paskievitch. S. Estes, and J.C. Lahr. Catalog of earthquake hypocenters for Augustine. Redoubt. Iliamna. and Mount Spurr volcanoes. Alaska: January 1. 1991 - December 31. 1993. USGS Open-File Report 96 70. p. 90. 1996.

Juhle. W.. and H. Coulter. The Mt. Spurr eruption. July 9. 1953. EOS Trans. 36. 199-202. 1955.

Julian. B.R.. Volcanic tremor: Nonlinear excitation by fluid flow. J. Geophys. Res.. 11859 11877.1994.

Keith T.E.C., J.M. Thompson, and R.G. McGimsey, Chemistry of crater lake waters prior to the 1992 eruptions of Crater Peak. Mount Spurr volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bull. 2139. 59-64. 1995.

Klein. F.W.. R.Y. Koyanagi. J.S. Nakata. and W.R. Tanigawa. The seismicity of Kilauea's magma system, in Volcanism in Hawaii, edited by R.VV. Decker. T.L. Wright, and P.H. Stauffer. U.S. Geol. Surv.. Prof. Pap.. 1350. 1019-1186. 1987.

Koyanagi. R.Y.. B A. Chouet. and K. Aki. Origin of volcanic tremor in Hawaii Part I. Data from Hawaii Volcano Observatory. 1969 - 1985. in Volcanism in Hawaii, edited by R. W. Decker. T.L. Wright, and P.H. Stauffer. U.S. Geol. Surv.. Prof. Pap.. 1350. 1221-1259. 1987.

Lahr. J.C.. HYPOELLIPSE/version 2.0. a computer program for determining local earthquake hvpocentral parameters, magnitude, and first motion. U.S. Geol. Surv. Open - File Report 89 116. p. 1002. 1989.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71

Lahr. J.C.. B.A. Chouet. C.D. Stephens. J.A. Power, and R.A. Page, Earthquake classification, location, and error analysis in a volcanic environment: implications for the magmatic system of the 1989 - 1990 eruptions of Redoubt Volcano. Alaska. J. Vole, and Geotherm. Res.. 62. 137-152. 1994.

Lee. W.H.K.. D.M. Tottingham. and J.O. Ellis. Design and implementation of a PC-based seismic data acquisition, processing, and analysis system, in Toolkit for seismic data acquisition, processing, and analysis, edited by W. H. K. Lee. El Cerito CA. Seismological Society of America. International Association of Seismology and Physics of the Earth's Intenor Software Library. 1. 21-46. 1989.

Malone. S.D.. Volcanic earthquakes: Examples from Mount St. Helens, in Earthquakes: Observations. Theroy, and Interpretation, edited by H. Kanamori and E. Boschi. Elsevier/North Holland. Amsterdam, 436-455, 1983.

Malone. S.D.. and S. Moran. Deep long-period earthquakes in the Washington Cascades. Eos Trans. 74. 438. 1997.

Mori. J.. R.A. White. D.H. Harlow. P. Okubo. J.A. Power. R.P. Hoblitt. E.P. Laguerta. A. Lanuza. and B.C. Bautista. Volcanic earthquakes following the 1991 climatic eruption of Mount Pinatubo: strong seismicity during a waning eruption, in Fire and Mud. eruptions and lahars of Mount Pinatubo. Philippines, edited by C. G. Newhail. and R. S. Punongbuyan. University of Washington Press. Seattle. WA. p. 339-350. 1996.

McNutt. S.R.. G. Tytgat. and J.A. Power. Preliminary analyses of volcanic tremor associated with 1992 eruptions of Crater Peak. Mount Spurr Volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bull. 2139. 161-177. 1995.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72

Neal. C.A.. R.G. McGimsey, C.A. Gardner, M.L. Harbin, and CJ. Nye. Tephra-fall deposits from the 1992 eruptions of Crater Peak. Mount Spurr Volcano. Alaska: a preliminary report on distribution, stratigraphy, and composition, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bull. 2139, 65-81. 1995.

Nye. C.J.. and D.L. Turner. Petrology, geochemistry, and age of the Spurr volcanic complex, eastern Aleutian arc. Bull Volcanol 52: 205-226. 1990.

Nye. CJ.. M.L. Harbin. T.P. Miller. S.E. Swanson, and C.A. Neal, Whole rock major- and trace-element chemistry of 1992 ejecta from Crater Peak. Mount Spurr Volcano. Alaska, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T. E. C. Keith. USGS Bull. 2139. 119-128, 1995.

Okada. H.. H. Watanabe. I. Yamahita. and I. Yokovama. Seismological significance of the 1977 - 1978 eruptions and magma intrusion process of Usu Volcano. Hokkaido, J. Vole, and Geotherm. Res.. 9.311-344.1981.

Pitt. A.M.. and D.P. Hiil. Long-period earthquakes in the Long Valley caidera region, eastern California. Geophys. Res. Let.. 21. 1679-1682. 1994.

Power. J.A.. Sesimicitv associated with the 1986 eruption of Augustine Volcano. Alaska. Masters Thesis. University of Alaska. Fairbanks. AK, p. 142. 1988.

Power. J.A.. A.D. Jolly, R.A. Page, and S.R. McNutt. Seismicity and forecasting of the 1992 eruptions of Crater Peak vent. Mount Spurr volcano. Alaska: an overview, in The 1992 eruptions of Crater Peak vent. Mount Spurr Volcano. Alaska, edited by T.E.C. Keith. USGS Bull. 2139. 149-159. 1995.

Power. J.A.. Lahr. J.C.. R.A. Page. B.A. Chouet. C.D. Stephens. D.H. Harlow. T.L. Murray, and J.N. Davies. Seismic evolution of the 1989 - 1990 eruption of Redoubt Volcano. Alaska. J. Vole, and Geotherm. Res.. 62. 69-94. 1994.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73

Power. J.A.. A. Villasenor. H.M. Benz. A seismic image of the Mount Spurr magmatic system. Bull. Volcanol.. 60. 27-37. 1998.

Riehle J. R.A reconnaissance of the major Holocene tephra deposits in the upper Cook Inlet region. Alaska. J. Vole, and Geotherm. Res. 26. 37-74. 1985.

Scandone. R.. and S.D. Malone. Magma supply, magma discharge and readjustment of the feeding system of Mount St. Helens during 1980. J. Geophys. Res.. 23. 239-262. 1985.

Turner. D.L.. and E.M. Wescott. Summary and conclusions of the Mount Spurr Alaska geothermal assessment project, in Geothermal energy resource investigations at Mount Spurr Alaska, edited by D. L. Turner and E. M. Wescott. University of Alaska Geophysical Institute Report UAG R-308.6.1 - 6.6. 1986.

Ukawa. M.. Deep low frequency earthquakes beneath volcanic front in central Japan. EOS Trans.. 78. 438. 1997

Ukawa. M.. and M. Ohtake. A monochromatic earthquake suggesting deep-seated magmatic activity beneath the Izu-Ooshima volcano. Japan. J. Geophys. Res.. 92. 12649-12663. 1987.

Walter. S.R.. Ten years of earthquakes at Lassen Peak. Mount Shasta, and Medicine Lakes volcanoes, nonhem California: 1981 - 1990. Seismol. Res. Let. 62. 25. 1991.

Wescott. E.M.. D.L.Tumer. C J. Nye. R.J. Motyka. and P. Moore. Exploration for geothermal energy resources at Mount Spurr. Alaska. Geotherm Resources Council Trans 12:203-210. 1988.

White. R.A.. Precursory deep long-period earthquakes at Mount Pinatubo: spatio-temporal link to a basalt trigger, in Fire and Mud. eruptions and lahars of Mount Pinatubo. Philippines.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74

edited by C.G. Newhali and R.S. Punongbuyan. University of Washington Press. Seattle. WA. p. 307-328. 1996

White, R.A. and D. Dzurisin. Deep long-period earthquakes: seismic evidence of rising basalt. EOS Trans. 78.437. 1997.

Wiemer. S. and S.R. McNutt. Variations in the frequency-magnitude distribution with depth in two volcanic areas: Mount St. Helens. Washington, and Mt. Spurr, Alaska, Geophys. Res. Let.. 24. 189-192. 1997.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4

Spatial Variations in the Frequency-Magnitude Distribution of Earthquakes at Soufriere Hills Volcano, Montserrat, West IndiesJ

4.0 Abstract The frequency - magnitude distribution of earthquakes measured by the 6-value is determined as a function of space beneath Soufriere Hills Volcano. Montserrat, from data recorded between August I. 1995 and March 31. 1996. A volume of anomalously high 6- values (6 > 3.0) with a 1.5 km radius is imaged at depths of 0 and 1.5 km beneath English's Crater and Chance's Peak. This high 6-value anomaly extends southwest to Gage's Soufriere. At depths greater than 2.5 km volumes of comparatively low 6-values (6-1) are found beneath St. George's Hill. Windy Hill, and below 2.5 km depth and to the south of English’s Crater. We speculate the depth of high 6-value anomalies under volcanoes may be a function of silica content, modified by some additional factors, with the most siliceous having these volumes that are highly fractured or contain high pore pressure at the shallowest depths.

4.1 Introduction The frequency of occurrence of earthquakes with increasing magnitude can be described as a power law by the equation,

/og/o/V = a - 6/Vf ( I )

where N is the cumulative number of earthquakes with magnitudes equal or larger to M. and a and 6 are constants (Ishimoto and Ida. 1939: Gutenberg and Richter. 1944). The 6-value is

'Published as: Power. J.A.. M. Wyss. and J.L. Latchman. Spatial Variations in the frequency-magnitude distribution of earthquakes at Soufriere Hills Volcano. Montserrat. West Indies. Geophysical Research Letters. 25. 3.653 - 3.656. 1998.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76

the slope of the best fit line between the observed number of earthquakes at a given magnitude and the magnitude. It is inversely proportional to the mean magnitude, thus differences in b reflect differences in the average crack length that ruptures in an earthquake. Values for b are generally close to 1 in much of the earth’s crust (Frohlich and Davis, 1994. Morgan et al.. 1988). However, in many volcanic areas b is frequently found to be much higher, often closer to 2 (e.g. Warren and Latham. 1970). Factors that can alter the b- value include increased heterogeneity of the material (Mogi. 1962), an increase in applied shear stress (Scholz. 1968: Urbancic et al.. 1992). an increase in the effective stress (Wyss. 1973). or an increase in the thermal gradient (Warren and Latham. 1970). In this study we examine spatial differences in relative ^-values at Soufriere Hills Volcano on Montserrat. West Indies, from data recorded between August I. 1995 and March 31, 1996 (Figure 4.1). Between August and October 1995 approximately 1,500 earthquakes were located beneath the volcano, likely associated with the accumulation of magma. These earthquakes covered a broad area of the southern portion of the island of Montserrat (Figure 4.2) (Aspinail et al.. in press). The most notable clusters of earthquake hypocenters occurred beneath English's Crater and in an elongate zone to the northeast extending from English’s Crater to the Bethel area (August 5 and 6). St. George's Hill (August 13 - 31. 1995), and Windy Hill (September August 25 - September 13) (for geographic locations see Figure 4.1). From October 1995 through March 1996 magma was actively erupting, forming a lava dome in English's Crater and seismic activity was more concentrated at shallow depth beneath the volcano (Miller et al.. in press. White et al.. in press), although the areas beneath St. George's Hill and Windy Hill remained active throughout the study period.

4.2 Data Beginning in late July 1995 the island of Montserrat was monitored by a network of 6 to 9 short period seismic stations (Figure 4.1). Signals from the various stations were telemetered to a central site and recorded on a PC-based seismic acquisition system similar to that described by Murray et al. (1996). Earthquake hypocenters and magnitudes were determined using the program HYP071PC (Lee and Valdes. 1989) and a velocity model which was derived for Montserrat by trial and error modifications to the velocity model used

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77

at Guadeloupe. West Indies (C. Antenor. pers. comm.. 1995). Magnitudes were determined by coda measurements using the relationship

Mb = 2.073*LOG(t) + 0.0018*D - 0.705 (2)

where t is the coda duration in seconds and D is the hypocentral distance in km (Sheperd and Aspinall. 1983). The period from August 1. 1995 to March 31. 1996 was selected for this study because the reporting is homogeneous, that is. no changes in magnitude of completeness and no inadvertent changes in magnitude scale, as often encountered in catalogs (Habermann 1982, Habermann 1991). occur in this period. After March 31. the catalog contains a much higher percentage of events without magnitude, such that these data were not useful for this study. During the study period, roughly 4.500 earthquakes were located. For approximately 200 of these earthquakes no coda duration was measured so no magnitude could be determined. An examination of FMD (Frequency Magnitude Distribution) plots suggests the catalog is complete above about magnitude 1.7. resulting in about 1900 earthquakes used for this analysis. The duration of some of the largest events exceeded the length of the trigger window on the event detection acquisition system, so we calculated magnitudes for these events using the regional Caribbean network, operated by the Seismic Research Unit of the University of West Indies. In total. 16 earthquakes were given a regionally determined magnitude. To maintain consistency, the locations for these events were calculated using only the short-period stations on Montserrat.

4.3 Method The method of 6-value estimation used here is identical to that described by Wiemer and Wyss (1997). Wiemer and Benoit (1996), and Wyss et al. (1997). The 6-value is estimated at the nodal points of a two-dimensional grid using the N nearest earthquakes. The nearest events are selected on the basis of epicenters in the case of map views, and on the basis of the nearest distance in the projections of the hypocenters onto a vertical plane in the case of cross sections. Consequently, sampling volumes have the shape of cylinders. The nodal separation we use is 0.0025 degrees (roughly 0.25 km) in maps and 0.25 km in cross

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78

sections, and N = 100. We restrict the radii of cylinders to a maximum value. , beyond which we do not use the 6-value as it is not based on sufficiently local data. Rm was set at 2.0 km. The 6-value is calculated using both the maximum likelihood method and the weighted least squares method. The use of both methods allows us to verify that the results are independent of the calculation method. The 6-value estimates shown in maps and cross sections are calculated with the maximum likelihood method.

4.4 Results: The most prominent anomaly is a volume of relatively high 6-values (red colors) which extends southwest from roughly Chance's Peak - English's Crater to the area of maximum 6-value anomaly near Gage's Soufrieres (both lower and upper)(Figure 4.2B). This high 6-value anomaly is imaged to a depth of about 1.5 km beneath Chance's Peak - English’s Crater and to 2.0 to 2.5 km beneath Gage’s Soufriere (Figure 4.2C). A strong zone of relatively low 6-value (blue colors) is imaged at depths of 3 to 5 km beneath St. George’s Hill and Windy Hill (Figures 4.2B and 4.2C). Another zone of low 6-values is seen from approximately 2-4 km depth beneath the volcanic edifice, as well as to the south of English’s Crater (Figures 4.2B - 4.2D). To demonstrate the difference in observed 6-values beneath the English's Crater - Chance's Peak - Gage's Soufriere ( "volcanic area") and St. George's Hill - Windy Hill (“outlying area"), we have compared FMDs using a 100 event sample in each area. The locations used for these circles are shown in Figure 4.2C. The volcanic area has a 6-value of 3.07 + 0.7. while in the outlying area b equals 0.92 + 0.03 (Figure 4.3). Both selections closely match the expected Gutenberg-Richter distribution. There is less than a 1 percent chance that the two distributions come from the same population of earthquakes (Utsu. 1992). To verify these results we inspected visually a number of FMDs calculated with 100 events each from the volcanic area with samples from the outlying area, as well as areas south of English’s Crater. None of these results varied greatly from the results shown in Figure 4.3. Based on this test and the spatial and temporal differences of the sample locations, we are confident that the observed distribution of 6-values at Montserrat is real and not the result of computational artifacts.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79

For this study we have used an Me of 1.7. It has been suggested that at many active volcanoes a bi-modal FMD may exist, perhaps as a result of earthquake families at shallow depth (Wiemer, written comm., 1997). Numerous families of similar earthquakes were observed at Montserrat (White et al.. in press), although the magnitudes are generally smaller than 1.7. Therefore, our study is not strongly affected by this problem.

4.5 Discussion and Conclusions The 6-values calculated for this study are on average higher than in most other areas, where they range from 0.5 to 1.5. This has been observed in some catalogs as well (Westerhaus, written comm. 1998) and may be a consequence of a compressed magnitude scale, possibly resulting from a number of assumptions we have made in computing magnitudes. Consequently, we consider the 6-values presented here to reflect relative values and use them only for comparative purposes at Montserrat. Our results indicate that the 6-value of earthquakes recorded during the study period shows significant spatial variation on the island of Montserrat. Specifically, the 6-value at roughly 0-2 km depth beneath English’s Crater - Chance's Peak - Gage's Soufriere was found to be much higher than in surrounding areas, especially beneath St. George's Hill and Windy Hill. Possible explanations for these high 6-values include increased heterogeneity, temperature, and stress conditions. AH of these effects are expected at 0 - 2 km depth beneath Chance’s Peak and English's Crater. During the early portion of the study period increased heat flow in the form of increased fumorolic activity was observed as weil as the formation of ground cracks (Aspinall et al., in press). During the later portions of the study period at both Chance’s Peak and English’s Crater increased heat flow and rock fracture are expected associated with the eruption of highly viscous andesitic magma. The Gage's Soufriere is an area where high heat flow would be expected as well as highly fractured and thermally altered material associated with the long-term geothermal activity observed there. There are also a number of possible explanations for the lower 6-values observed beneath St. George’s Hill and Windy Hill, and to the south of English’s Crater. We speculate that seismicity here might have been triggered by stress generated by the intrusion of magma beneath English's Crater. We think these areas are likely less fractured and have less thermal alteration than those beneath English’s Crater and Gage's Soufriere. These areas of more

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80

competent rock could support larger earthquakes, which is reflected in the lower observed 6- values. To date, detailed analyses of spatial variation in the FMD have been completed at Mount St. Helens and Mount Spurr in the United States (Wiemer and McNutt. 1997) and off- Ito in Japan (Wyss et al.. 1997). At off-Ito. high 6-regions were found to reflect highly fractured conditions surrounding magma chambers while "normal” 6-values less than 1 were found surrounding these areas. Wyss et al. (1997) suggested that even though basaltic magma ascended through these normal areas, the size of the dike systems were likely too small to be imaged by this technique. We did not attempt to examine possible changes in 6-values at Montserrat as a function of time, because the catalog was found to be inhomogeneous with respect to magnitude after March 1996. Observations presented by Miller et al. (in press) indicate that much larger earthquakes were recorded at shallow depth beneath English’s Crater during later periods of the eruption. We believe that analysis of 6-values as a function of space and time holds much promise in constraining volcanic processes at active volcanoes. As at other volcanoes, where 6-values have been imaged using this technique, high 6- value anomalies at Montserrat are confined to small volumes with radii of about 1.5 km. However at Montserrat the high 6-values are seen at very shallow depth (0-2 km), whereas the shallowest anomaly at Mount St. Helens is located at 3 km. at Mount Spurr at 4 km. and at off-Ito volcano below 7 km. A possible explanation is that the relatively viscous andesite (60 Vc Si0:) (Devine et al.. in press) at Montserrat fractures the volcanic edifice to a much higher degree than the basalt at off-Ito. or the basaltic-andesite at Mount Spurr. The exception at Mount St. Helens where a dacite magma (63-65 °7c SiO:) erupts to form a lava dome, may be that Wiemer and McNutt (1997) examined data from 1988 through 1996. a period when Mount St. Helens was not in active eruption.

4.6 Acknowledgments This study would not have been possible without the great efforts and long hours worked by many people associated with MVO during 1995 and 1996. The assistance provided by Stefan Wiemer with the ZMAP software is gratefully acknowledged. This project was supported by funds from the US Geological Survey Volcano Hazards Program, the University of Alaska

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81

Geophysical Institute, the National Science Foundation through grant EAR-9614783, and the Wadati Foundation at the Geophysical Institute. University of Alaska. Fairbanks.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82

Figure 4.1. Montserrat seismic network. July 1995 - March 1996: Montserrat seismic network with individual stations shown as triangles from July 1995 to March 1996. Letters correspond to approximate geographic locations referenced in text: CE - Chance’s Peak/English Crater. GS - Gage s Soufriere (upper and lower), SGH - St. George’s Hill, WH - Windy Hill. B - Bethel. ~

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83

12' 6 Z W ir

Figure 4.2. Maps and cross-sections showing b-values and earthquakes used in this study: A) Map of the southern portion of the island of Montserrat showing earthquake epicenters used in this study, seismic stations, and the location of cross-section E-E’ (Figure 4.2C). B) Map view of 6-values for earthquakes above 2.5 km depth. C) Cross section E-Er showing the distribution of 6-values. Circles show locations used to calculate FMDs shown in figure 4.3. D) Map view of 6-values calculated for earthquakes at greater than 2.5 km depth. X denotes location of largest earthquake in B and D.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. S4

Figure 4.3. Frequency-magnitude distnbutions from English’s Crater and St. Georges Hill: Frequency magnitude distributions from the English's Crater- Chance's Peak - Gage's Soufriere area (A) and from the St. George's Hill - Windy Hill area

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. S5

4.7 References

Aspinall. W.. A.D. Miller. L. L. Lynch. J.L. Latchman. R.A. Stewart. J.A. Power, and MVO staff. Soufriere Hills eruption. Montserrat. 1995 - 1997. volcanic earthquake locations and fault plane solutions. Geophy. Res. Lett., in press.

Devine, J.D. M.D. Murphy, M. J. Rutherford. J. Barclay. R.S.J. Sparks. M.R. Carroll. S.R. Young, J.E. Gardner. Petrologic evidence for preemptive pressure-temperature conditions, and recent reheating, of andesitic magma erupting at the Soufriere Hills Volcano. Montserrat. W.I. Geophy. Res. Lett., in press.

Frohlich, C.. and S. Davis, Teleseismic b- values: Or much ado about 1.0, J Geophys. Res.. 98. 631-634, 1994.

Gutenberg, B, and C.F. Richter. Frequency of Earthquakes in California. Bull. Seism., Soc. Am.. 34. 185-188. 1944.

Habermann. R.E.. Consistency of teleseismic reporting since 1963. Bull. Seism. Soc. Am.. 72. 93-112. 1982

Habermann. R.E.. Seismicity rate variations and systematic changes in magnitudes in teleseismic catalogs. Techtonophysics, 193. 277-289. 1991.

Ishimoto. M.. and K. Iida. Observations of earthquakes registered with the microseismograph constructed recently. Bull. Earthq. Res. Inst.. 17. 443-478. 1939.

Lee. W.H.K.. C.M. Valdes. User Manual for HYP071PC. in IASPEI Software Library. Volume I. Toolbox for Seismic Data Acquisition, Processing, and Analysis. 1989.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86

Miller. A.D. R.S. Stewart, L.L. Lynch, J.L. W.P. Aspinall. R.A. White. Seismicity associated with dome growth and collapse at Soufriere Hills Volcano. Montserrat. Geophy. Res. Lett., in press.

Mogi. K... Magnitude-frequency relation for elastic shocks accompanying fractures of various materials and some related problems in earthquakes. Bull. Earthq. Res. Inst.. 40. 831-853, 1962.

Morgan. F. D.. G. Wadge. J. Latchman, W.P. Aspinall. D. Hudson, and F. Samstag, The earthquake hazard alert of September 1982 in southern Tobago. Bull. Seis. Soc. Am. 78, 1550-1562. 1988.

Murray, T.L.. J.A. Power. G. Davidson, and J.N. Marso, A PC-based real-time volcano data- acquisition and analysis system, In (editors) C.G. Newhall and R.S. Punongbayan, Fire and Mud, Eruptions and Lahars of Mount Pinatubo, Philippines. University of Washington Press, p 225-232. 1996.

Scholz. C.H.. The frequency-magnitude relation of microfracturing in rock and its relation to earthquakes. Bull. Seism. Soc. Am.. 58. 399-415. 1968.

Sheperd. J.B.. and W'.P. Aspinall. Seismicity and earthquake hazard in Trinidad and Tobago. West Indies, International Journal of Earthquake Engineering and Structural Dynamics. Vol. 11.229 -250. 1983.

Urbancic. T.I.. C.I. Trifu. J.M. Long, and R.P. Young. Space-time correlations of ^-values with stress release. Pageoph. 139. 449-462. 1992.

Utsu. T.. On seismicity, in: Report of Cooperative Research of the Institute of Statistical Mathematics 34. Mathematical Seismology VII. Annals of the Institute of Statistical Mathematics. Tokyo. 139-157. 1992.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87

Warren. N.W.. and G.V. Latham. An experimental study of thermally induced microfracturing and its relation to volcanic seismicity. J. Geophys. Res.. 75. 4455-4464. 1970.

White. R.A.. A.D. Miller. W. P. Aspinal. L.L. Lynch. J.A. Power, Observations of hybrid seismic events at Soufriere Hills volcano. Montserrat. July 18 1995 to October 1997. G.R.L, in press.

Wiemer, S.. and J. Benoit. Mapping the 6-value anomaly at 100 km depth in the Alaska and New Zealand subduction zones. Geophys. Res. Letters. 23, 1557-1560. 1996.

Wiemer, S.. and McNutt. S. R.. Variations in the frequency-magnitude distribution with depth in two volcanic areas: Mount St. Helens, Washington, and Mount Spurr Alaska. Geophys. Res. Let.. 24. 189-192. 1997.

Wiemer, S.. and M. Wyss, Mapping the frequency-magnitude distribution in asperities: an improved technique to calculate recurrence times? Jour. Geophys. Res.. 102. 15.115-15.128. 1997.

Wyss. M.. Towards a physical understanding of the earthquake frequency distribution. Geophys. J. R. Astr. Soc.. 31. 341-359. 1973.

Wyss, M.. K. Shimazaki. and S. W'iemer. Mapping active magma chambers by 6-values beneath off-Ito Volcano. Japan. J. Geophys. Res.. 102. 20,413-20.433. 1997.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IMAGE EVALUATION

1.0 y£ ... 13J2 l-i tm m ===== m £ 11 ui: la i i a

m 1.25 1 1.6

150m m

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.