USF Libraries
USF Digital Collections

Seismic and geodetic investigation of the 1996-1998 earthquake swarm at strandline lake, alaska

MISSING IMAGE

Material Information

Title:
Seismic and geodetic investigation of the 1996-1998 earthquake swarm at strandline lake, alaska
Physical Description:
Book
Language:
English
Creator:
Kilgore, Wayne
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Velocity model
B-value
Relocation
InSAR
Volcanic arc
Crustal deformation
Dissertations, Academic -- Geology -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Microearthquake (< M3.0) swarms occur frequently in volcanic environments, but do not always culminate in an eruption. Such non-eruptive swarms may be caused by stresses induced by magma intrusion, hydrothermal fluid circulation, or regional tectonic processes, such as slow-slip earthquakes. Strandline Lake, located 30 km northeast of Mount Spurr volcano in south-central Alaska, experienced an intense earthquake swarm between August 1996 and August 1998. The Alaska Volcano Observatory (AVO) catalog indicates that a total of 2,999 earthquakes were detected during the swarm period, with a maximum magnitude of Mw 3.1 and a depth range of 0-30 km below sea level (with the majority of catalog hypocenters located between 5-10 km BSL). The cumulative seismic moment of the swarm was 2.03e15 N-m, equivalent to a cumulative magnitude of Mw 4.2. Because of the swarm's distance from the nearest Holocene volcanic vent, seismic monitoring was poor and gas and GPS data do not exist for the swarm period. However, combined waveforms from a dense seismic network on Mount Spurr and from several regional seismic stations allow reanalysis of the swarm earthquakes. I first developed a new 1-D velocity model for the Strandline Lake region by re-picking and inverting precise arrival times for 27 large Strandline Lake earthquakes. The new velocity model reduced the average RMS for these earthquakes from 0.16 to 0.11s, and the average horizontal and vertical location errors from 3.3 to 2.5 km and 4.7 to 3.0 km, respectively. Depths of the 27 earthquakes ranged from 10.5 to 22.1 km with an average depth of 16.6 km. A moderately high b-value of 1.33 was determined for the swarm period, possibly indicative of magmatic activity. However, a similarly high b-value of 1.25 was calculated for the background period. 28 well-constrained fault plane solutions for both swarm and background earthquakes indicate a diverse mixture of strike-slip, dip-slip, and reverse faulting beneath Strandline Lake. Finally, five Interferometric Synthetic Aperture Radar (InSAR) images spanning the swarm period unambiguously show no evidence of surface deformation. While a shallow volcanic intrusion appears to be an unlikely cause of the Strandline Lake swarm based on the new well-constrained earthquake depths and the absence of strong surface deformation, the depth range of 10.5 to 22.1 km BSL for relocated earthquakes and the high degree of FPS heterogeneity for this swarm are similar to an earthquake swarm beneath Lake Tahoe, California in 2003 caused by a deep intrusion near the base of the crust (Smith et al, 2004, Science 305, 1277-1280). This similarity suggests that a deep crustal magmatic intrusion could have occurred beneath the Strandline Lake area in 1996-1998 and may have been responsible for the resulting microearthquake activity.
Thesis:
Thesis (M.S.)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Wayne Kilgore.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0003487
usfldc handle - e14.3487
System ID:
SFS0027802:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003487
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Kilgore, Wayne.
0 245
Seismic and geodetic investigation of the 1996-1998 earthquake swarm at strandline lake, alaska
h [electronic resource] /
by Wayne Kilgore.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (M.S.)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Microearthquake (< M3.0) swarms occur frequently in volcanic environments, but do not always culminate in an eruption. Such non-eruptive swarms may be caused by stresses induced by magma intrusion, hydrothermal fluid circulation, or regional tectonic processes, such as slow-slip earthquakes. Strandline Lake, located 30 km northeast of Mount Spurr volcano in south-central Alaska, experienced an intense earthquake swarm between August 1996 and August 1998. The Alaska Volcano Observatory (AVO) catalog indicates that a total of 2,999 earthquakes were detected during the swarm period, with a maximum magnitude of Mw 3.1 and a depth range of 0-30 km below sea level (with the majority of catalog hypocenters located between 5-10 km BSL). The cumulative seismic moment of the swarm was 2.03e15 N-m, equivalent to a cumulative magnitude of Mw 4.2. Because of the swarm's distance from the nearest Holocene volcanic vent, seismic monitoring was poor and gas and GPS data do not exist for the swarm period. However, combined waveforms from a dense seismic network on Mount Spurr and from several regional seismic stations allow reanalysis of the swarm earthquakes. I first developed a new 1-D velocity model for the Strandline Lake region by re-picking and inverting precise arrival times for 27 large Strandline Lake earthquakes. The new velocity model reduced the average RMS for these earthquakes from 0.16 to 0.11s, and the average horizontal and vertical location errors from 3.3 to 2.5 km and 4.7 to 3.0 km, respectively. Depths of the 27 earthquakes ranged from 10.5 to 22.1 km with an average depth of 16.6 km. A moderately high b-value of 1.33 was determined for the swarm period, possibly indicative of magmatic activity. However, a similarly high b-value of 1.25 was calculated for the background period. 28 well-constrained fault plane solutions for both swarm and background earthquakes indicate a diverse mixture of strike-slip, dip-slip, and reverse faulting beneath Strandline Lake. Finally, five Interferometric Synthetic Aperture Radar (InSAR) images spanning the swarm period unambiguously show no evidence of surface deformation. While a shallow volcanic intrusion appears to be an unlikely cause of the Strandline Lake swarm based on the new well-constrained earthquake depths and the absence of strong surface deformation, the depth range of 10.5 to 22.1 km BSL for relocated earthquakes and the high degree of FPS heterogeneity for this swarm are similar to an earthquake swarm beneath Lake Tahoe, California in 2003 caused by a deep intrusion near the base of the crust (Smith et al, 2004, Science 305, 1277-1280). This similarity suggests that a deep crustal magmatic intrusion could have occurred beneath the Strandline Lake area in 1996-1998 and may have been responsible for the resulting microearthquake activity.
590
Advisor: Diana Roman, Ph.D.
653
Velocity model
B-value
Relocation
InSAR
Volcanic arc
Crustal deformation
690
Dissertations, Academic
z USF
x Geology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.3487



PAGE 1

Seismic and Geodetic Investigation of the 1996 1998 Earthquake Swarm at Strandline Lake, Alaska b y Wayne Walter Kilgore A thesis submitted in partial fulfillment of the requirement s for the degree of Master of Science Department of Geology College of Arts and Sciences University of South Florida Major Professor: Diana Roman, Ph.D. Sarah Kruse, Ph.D. Paul Wetmore, Ph.D. Date of Approval: April 15 2010 Keywords: velocity model, focal mechanism, b value, relocation, InSAR, volcanic arc, crustal deformation Copyright 2010, Wayne Walter Kilgore

PAGE 2

A cknowledgements First and f oremost I would like to convey my deepest gratitude to my advisor, Dr. Diana Roman, without whose unfailing patience, instruction, and encouragement this work would not have come to fruition Throughout my graduate studies, no other person has been more influential and engaging than Dr. Roman. One simply could not wish for a better advisor. Much of the data presented in this thesis is a result of the collaborative effort between me and several researchers from other institutes. I would like to give a special thanks to Dr. John Power and rest of the staff at the Alaska Volcano Observatory for their assistance and insight s during my data gathering and a nalysis. Thank you to Dr. Roger Hansen and the Alaska Earthquake Information Center for the additional earthquake data, which played a major part in the completion of this study. Likewise, I want to express my appreciation to Dr. Juliet Biggs and the Ros enstiel School of Marine and Atmospheric Science for their guidance in gathering InSAR images for the deformation analysis. Furthermore, I would also like to thank my committee members, Dr. S arah Kruse and Dr. Paul Wetmore for the time and effort they spent in reviewing and editing my thesis. I am also grateful to the other members of the volcano seismology laboratory , and Mel Ro dgers, and m y other colleagues in the USF geology department who have supported me during this e ntire process. To all of you, my most heartfelt gratitude.

PAGE 3

i Table of Contents List of Figures i i i List of Tables iv Abstract v i Chapter 1: Introduction 1 1.1. Motivation 4 Chapter 2: Background 6 2.1 Earthquake Swarms 7 2.1.1. Volcano Tectonic Earthquake Swarms 8 2.1.2. Non Volcanic Earthquake Swarms 11 2.2 Strandline Lake, Alaska 1 2 2.2.1. Physiographic and Geologic Setting 14 2.2.2. Tectonic Setting 16 2 .2.3. Geophysical Monitoring 17 2.2.4. The 1996 1998 Earthquake Swarm 19 2.2.4.1. Previous Work 2 3 Chapter 3: Methods 25 3.1 B values 26 3.2. Velocity Model 27 3.3 Hypocenter Relocations 32 3.4 Fault Plane Solution Analysis 3 4 3.5 Surface Deformation 3 6 Chapter 4: Results 38 4.1. B values 39 4.2. 1 D Velocity Model 41 4.3. Earthquake Locations 50 4.4. Stress Analysis 53 4.5. InSAR Analysis 58 Chapter 5: Discussion 60 Chapter 6: Conclusions 6 8

PAGE 4

ii List of References 7 1 Appendices 75 Appendix A I n SAR images of Strandline Lake 76

PAGE 5

iii List of Figures Figure 1.1 Location m ap of the Cook Inlet region of Alaska 2 Figure 1.2 Tectonic map of southern Alaska 5 Figure 2.1 Example s of V T seismicity 9 Figure 2.2 Existing models showing the relationship between the propagation of magma through a dike and the observed VT seismicity 9 Figure 2.3 Geologic map of the Strandline Lake, Alaska area 15 Figure 2.4 Seismic networ k surrounding Strandline Lake 18 Figure 2.5 Plot showing the catalog depths of the swarm earthquakes over time 20 Figure 2.6 Histogram of the seismic activity at Strandline Lake, Alaska over a 10 year period 20 Figure 2.7 Plot of earthquake magnitudes over time for earthquak es at Strandline Lake 21 Figure 2.8 Plot showing the cumulative release of seismic energy from the earthquake swarm at Strandline Lake 2 1 Figure 3.1 Diagram of the Mount Spurr and Generic Alaska velocity models 29 Figure 3.2 Example earthquake seismograms from Strandline Lake 30 Figure 3.3 Illustration showing the principles of SAR imagery 37 Figure 4.1 Strandline Lake b values 40 Figure 4.2 Diagram of the Strandline Lake velocity model 42 Figure 4. 3 Map showing the locations and 95% confidence error ellipses of the 28 large Strandline Lake earthquakes obtained using the Generic Alaska velocity model 45

PAGE 6

iv Figure 4. 4 Map showing the locations and 95% confidence error ellipses of the 28 large Str andline Lake earthquakes obtained using the Mount Spurr velocity model 46 Figure 4. 5 Map showing the locations and 95% confidence error ellipses of the 28 large Strandline Lake earthquakes obtained using the Strandline Lake velocity model 47 Figure 4 6 North South cross section s of the 28 large Stran dline Lake earthquakes 48 Figure 4 7 Epicenters of Strandline Lake earthquakes using HYPOELLIPSE 5 1 Figure 4.8 North South cross section of the relocated earthquakes using the Strandline Lake model, corresponding to Figure 4.6b 52 Figure 4.9 Map of the Strandline Lake region showing focal mechanisms of eight large earthquakes recorded during the background period 54 Figure 4.10 Rose diagrams and stereonet plots of the P and T axes of the eight large background earthquakes at Strandline Lake 55 Figure 4.11 Map showing the FPS of the 20 large Strandline Lake earthquakes which occurred during the swarm period 56 Figure 4.12 Rose diagrams and stereonet plots of the P and T axes of the twenty large swarm earthquakes at Strandline Lake 57 Figure 4.13 Example InSAR image of Strandline Lake, Alaska 59 Figure 5.1 Map view and cross section from McNutt and Marzocchi (2004) showing seismic activity at Strandline Lake between October 1989 and December 2001 62 Figure 5.2 Rose dia gram from Wiemer et al., ( 1999 ) 67 Figure 5.3 Rose diagram compiled from P axis orientations given by Jolly and Page (1994) for tectonic earthquakes around Mount Spurr 67

PAGE 7

v List of Tables Table 3.1 The four layer Mount Spurr 1 D velocity model developed by Jolly and Page (1994) 28 Table 3.2 The nine layer generic Alaska 1 D velocity model 28 Table 4.1 New seven layer 1 D velocity model for the Strandline Lake area 42 Table 4.2 Station Corrections for the S trandline Lake velocity model 44 Table 4.3 Image pairs of the 13 InSAR images of Strandline Lake 59 Table 5.1 Comparison of results and statistics of the 1996 1998 swarm at Strandline Lake, Alaska to six other earthquake swarms which occurred in non volcanic areas 64

PAGE 8

vi Seismic and Geodetic Investigation of the 1996 1998 Earthquake Swarm at Strandline La ke, Alaska Wayne Walter Kilgore ABSTRACT Microearthquake (< M3.0) swarms occur frequently in volcanic environments, but do not always culminate in an eruption. Such non eruptive swarms may be caused by stresses induced by magma intrusion, hydrothermal fluid circulation, or regional tectonic processes, such as slow slip earthquakes. Strandline Lake, located 30 km northeast of Mount Spurr volcano in south cent ral Alaska, experienced an intense earthquake swarm between August 1996 and August 1998. The Alas ka Volcano Observatory (AVO) catalog indicates that a total of 2,999 earthquakes were detected during the swarm period, with a maximum magnitude of M w 3.1 and a depth range of 0 30 km below sea level (with the majority of catalog hypocenters located betwee n 5 10 km BSL). The cumulative seismic moment of the swarm was 2.03e 15 N m, equivalent to a cumulative magnitude of M w 4.2. Because of the swarm's distance from the nearest Holocene volcanic vent, seismic monitoring was poor and gas and GPS data do not exi st for the swarm period However, combined waveforms from a dense seismic network on Mount Spurr and from several regional seismic stations allow reanalysis of the swarm earthquakes. I first developed a new 1 D velocity model for the Strandline Lake region by re picking and inverting precise arrival times for 27 large Strandline Lake earthquakes. The new velocity model reduced the average RMS for these earthquakes from 0.16 to 0.11s, and the average

PAGE 9

vii horizontal and vertical location errors from 3.3 to 2.5 km and 4.7 to 3.0 km, respectively. Depths of the 27 earthquakes ranged from 10.5 to 22.1 km with an average depth of 16.6 km. A moderately high b value of 1.33 was determined for the swarm period, possibly indicative of magmatic activity. However, a similar ly high b value of 1.25 was calculated for the background period. 28 well constrained fault plane solutions for both swarm and background earthquakes indicate a diverse mixture of strike slip, dip slip, and reverse faulting beneath Strandline Lake. Finally five Interferometric Synthetic Aperture Radar (InSAR) images spanning the swarm period unambiguously show no evidence of surface deformation. While a shallow volcanic intrusion appears to be an unlikely cause of the Strandline Lake swarm based on the new well constrained earthquake depths and the absence of strong surface deformation, the depth range of 10.5 to 22.1 km BSL for relocated earthquakes and the high degree of FPS heterogeneity for this swarm are similar to an earthquake swarm beneath Lake Tahoe, California in 2003 caused by a deep intrusion near the base of the crust (Smith et al, 2004). This similarity suggests that a deep crustal magmatic intrusion could have occurred beneath the Strandline Lake area in 1996 1998 and may have been respons ible for the resulting microearthquake activity.

PAGE 10

1 Chapter 1 Introduction Earthquake swarms consist of spatiall y and temporally clustered micro earthquake s which lack the typi cal mainshock/aftershock sequence commonly associated with large crustal earthquakes. The cause of such swarms is a topic of debate within the geologic community (e.g. Vidale and Shearer, 2006) Some swarms are the seismic response to an injectio n of magmatic material, such as at Lake Tahoe, California in 2003 (Smith et al., 2004) and Upptyppingar, Iceland in 2007 (Jak obsdottir et al., 2008). Other earthquake swarms appear to have non volcanic origins and that the lack of a mainshock at the onset of the sequence does not necessarily denote the involvement of magma ( e.g. Jones and Malone, 2005). Both of these mechanisms are possible, and only by examining multiple pieces of evidence such as seismic, deformation, and gas emission data, can the spe cific cause of a swarm be determined. Strandline Lake is located in south central Alaska and lies approximately 110 km west northwest of Anchorage (Figure 1.1). The Strandline Lake area experienced an intense earthquake swarm between mid 1996 and mid 1998. The swarm consisted of approximately 3,000 events with a maximum magnitude of M w 3.1 and a depth range of approximately 3 to 12 km (McNutt and Marzocchi, 2004) below sea level (BSL ) In the peer reviewed literature, one study (Jolly and Page, 2004) describes earthquake activity at Strandline Lake as spatially distinct from other, more seismically

PAGE 11

2 Figure 1 .1: (purple star) proximity to and alignment with several volcanoes (red triangles). Black diamonds represent seismic stations and black dot s represent population centers.

PAGE 12

3 active areas around the C ook Inlet but concedes that the Strandline Lake events are poorly located due to sparse network of surrounding seismic monitoring stations. Here, data resulting from a merged seismic network will be analyzed in an attempt to prov ide well constrained locations and fault plane solutions for the Strandline Lake events. I t is hypothesize d that the earthquake swarm was caused by one of three mechanisms: 1) emplacement of a magma body such as a dike or sill, 2) microseismicity purel y tectonic processes, such as a slow slip event along the Aleutian Arc or 3) an increase in (non magmatic) fluid circulation In an attempt to determine the cause of the Strandline Lake earthquake swarm, several data sets and analy tical methods are employed. First, a b value is determined for both background and swa rm periods. Second, a 1 D velocity model is developed that is specific to the Strandline Lake area Third, using the new velocity model, the entire Alaska Volcano Observatory ( AVO ) cat alog of earthquakes in the Strandline Lake area i s relocated. Fourth, Synthetic Ap erture Radar (InSAR) images are analyzed to determine whether any localized uplift in response to an intrusion of a magmatic dike or sill can be detected Fifth, focal mech anisms of backg round and swarm earthquakes are examined in an attempt to detect any t emporal changes in stress at Strandline Lake. Sixth, the results of the aforementioned analyses are compared to published analyses of several other earthquak e swarms that occurred in a non volcanic setting. From these results, the most logical explanation for the source of the earthquake swarm at Strandline Lake will be determined.

PAGE 13

4 1.1 Motivation Earthquake swarms are a common occurrence worldwide; however, the swarm at Strandline Lake was chosen for in depth analyses for several reasons. First, Strandline Lake sits in a unique region at the northern edge of the Aleutian volcanic arc (Figure 1.1) at the point where the North American, Pacific plates, and Ya kutat terrane converge (Figure 1.2) Thus, the swarm could be a result of tectonic processes in conjunction with fluid upwelling from the edge of the subducting slab However, because it is located directly in line with the Aleutian volcanic front, the S trandline Lake area is in the correct location to be a developing cryptovolcano Second, no gas or GPS monitoring was conducted during the swarm period, making analysis of the seismicity the primary investigative tool. A unique velocity model for use i n locating earthquakes around the Strandline Lake area did not exist prior to this research Its development i s critical for the seismicity analysis and for future research in the Strandline Lake region Third, little knowledge currently exists concerning this particular swarm and most explanations of its cause are speculative. Only one study (McNutt and Marzocchi, 2004) attempts to hypothesize a cause for the swarm but this hypothesis was not rigorously tested Because Strand line Lake may reside in a transitional stress regime between the Aleutian Arc and the Alaska Range, understanding the cause of this earthquake swarm may be helpful in un derstanding the mechanisms that drive similar swarms in non volcanic regions around the world and Alaskan tectonics in general An accurate interpret ation of the driving process for this earthquake swarm may assist in forecasting future v olcanic eruptions and mitigating the associated natural hazards posed to surrounding communities.

PAGE 14

5 Figure 1.2 Tectonic map of southern Alaska. Strandline Lake (dark purple star) sits near the junction of the Pacific Plate, North American Plate, and the Yakutat block Volcanoes are represented by red triangles, depth to the top of the subducting slab by dashed blue lines, and major faults by solid green lines. (Figure after Eberhart Phillips, 2006) Hayes Redoubt Iliamna Spurr

PAGE 15

6 Chapter 2 Background The aim of this chapter is to provide a detailed explanation of current know ledge and unanswered questions about earthquake swarms, and about the Strandline Lake earthquake swarm in particular A detailed description of the geographic and tectonic setting and the seismicity of Strandline Lake, as well as the tools and methods us ed to monitor the area are fundamental to this understanding. Additionally, general characteristics of earthquake swarm activity, and particularly that which occurs in non volcanic areas, need to be examined before any interpretations of the Strandline Lake swarm can be made

PAGE 16

7 2.1 Earthquake Swarms Earthqua ke swarms are characterized as spatio temporal clusters of seismic events whi ch lack a dominant earthquake at the onset of the sequence (Fischer and Horalek, 2005). The number of earthquakes that c onstitute a swarm can range from tens to hundreds of thousands, yet swarm earthquakes generally have a maximum magnitude of less than 4 (Ma and Ea ton, 2009). Earthquake s warms are common occurrences in volcanic regions, in which case they are referred to as volcanotectonic (VT) earthquake swarms (Lahr et al., 1994), however, many have also been recorded in non volcanic areas ( Cappa et al., 2009; Ma and Eaton, 2009; Jakobsdottir et al., 2008; Von Seggern et al., 2008; Ibs von Seht et al., 2006; Vidale and Shearer, 2006; Fischer and Horalek, 2005; McNutt and Marzocchi, 2004; Smith et al., 2004; Chiu et al., 1984 ) While swarms are clearly observed in the seismic record their source mechanisms are complex and poorly und erstood. Earthquake swarms are thought to be caused by one or more of the following processes: T ectonic activity leading to a critical change in stress of the region intrusion of highly pressurized crustal fluids (including magma) or aseismic slip along a fault plane (Fischer and Horalek, 2005; Vidale and Shearer, 2006).

PAGE 17

8 2.1.1 Volcano Tectonic Earthquake Swarms Earthquake swarms are observed predominantly in volcanic environments and these VT swarms are one of the earliest detectable precursors to volcanic unrest (Roman and Cashman, 2006). seismicity within these VT swarms (Figure 2.1) is attributed to the brittle failure of rock Several conceptual models for the occurrence o f VT swarms have been proposed Hill (1977) first proposed that VT swarms are the result of a series of inflated dikes with their long axis orientated in th e direction of maximum principal stress. Hill (1977) made field observations that dikes commonly exhibited an en echelon pattern and proposed that offset dikes were connected by shear fractures and that VT seismicity occurred along these fractures (Figure 2.2A). Ukawa and Tsukahara (1996) suggested a diffe rent model based upon the seismicity distribution of several earthquake swarms within a monogenetic volcanic field off the east coast of the Izu Peninsula in central Japan. They proposed a spatio temporal relationship between hypocentral depths of the VT earthquakes and dike propagation. The VT seismicity is focused in a zone at the t op of the dike as tensile stress is elevated when the adjacent tip of the dike opens (Figure 2.2B). A third model proposed by Roman (2005) suggests that VT seismicity is foc used laterally outward from the dike as its walls are pushed outward during inflation. Notable characteristics of this model are the random spatial distribution of VT seismicity surrounding the dike as well as an approximate 90 rotation of VT fault plane solution p axe s relative to regional maximum compression (Figure 2.2C).

PAGE 18

9 Figure 2 .1 Examples of VT seismicity at Kuchinoerabujima Volcano, Japan (Figure from Triastuty et al., 2009). Figure 2.2 Existing models showing the relationship between the propagation of a magma filled dike and the observed VT seismicity. (A) Map view of t he model by Hill (1977) shows VT earthquakes along a shear plane extending from the tips of dikes. (B) Cross section of the model by Ukawa and Tsukahara (1996) shows a progressive upward migration of VT seismicity ahead of the tip of the inflating dike. (C) Cross section of the model proposed by Roman (2005) shows a random temporal and spatial distribution of the VT s eismicity along with a 90 r otation of the fault plane solution as th e dike inflates (Figure from Roman and Cashman 200 6 ).

PAGE 19

10 None of the three models mentioned can explain VT seismicity at all volca nic areas (Roman and Cashman, 2006). Instead, each model appears to address magma ind uced stresses in different areas surrounding a propagating or inflating dike. However, taken together, they provide a combined model for VT swarm s Roman and Cashman (2006) observed that the models supporting VT seismicity associated with dike propagatio n (e.g. Hill, 1977 and Ukawa and Tsukhara, 1996) are consistent with VT swarms at volcanoes with a basaltic composition while VT seismicity associated with dike inflation (e.g. Roman, 2005) occurred more commonly at stratovolcanoes with a more silicic magma compositio n. They concluded that mechanisms of VT seismicity are most likely related to pre existing structural features, the orientation and magnitude of regional stress, and the rheological properties of the ascending magma

PAGE 20

11 2.1.2 Non Volcanic Earthquake Swarms While most common in volcanic areas and continental rift zones earthquake swarms can also occur in areas without a n active volcano ( e.g., Brauer et al., 2009). Seismic activity in these areas generally result s from failure along pre existing fault planes and typically exhibit s mainshock aftershock sequences characterized by a rate of decay defined by the Omori Law (Hainzl and Ogata, 2005 Utsu, 1961 ). However, non volcanic swarms which deviate from Omori type behavior may occur when either magmatic or hydrothermal activity induce hundreds to thousands of small events with no precursory mainshock. A survey of previous studies of earthquake swarms shows that swarms in volcanic regions tend to be shallower than swarms in regions where volcanic activity is absent ( Jakobsdottir et al., 2008; von Seggern et al., 2008; Ibs von Seht, 2006; Vidale and Shearer, 2006; Jones and Malone, 2005; McNutt and Marzocchi, 2004; Roman et al., 2004; Smith et al., 2004 ). In examining these studies, i t appears that volcanic swarm activity is generally concentrated between ~10km and the vent of the volcanic edifice while non volcanic swarms occur near the base of the seismogenic crust It should be noted however that two of the examined studies of non volcanic swarms (Ma and Eaton, 2009; Cappa et al., 2009) document shallow ( < 10km ) hypocentral depths. Both of these swarms may have also been induced by the circulation of hydrothermal fluids within the shallow crust. Ma and Eaton (2009) stated that fluids play a critical role in swarm seismogenesis and that some swarms may be triggered by anthropogenic activities, such as the injection or removal of fluids ( e.g. Ake et al., 2005)

PAGE 21

12 2.2 Strandline Lake, Alaska Strandline Lake is a small glacially dammed lake in south central Alaska and was the site of an intense earthquake swarm beginning in mid 1996 and continuing until mid 1998. The following sections will give a compre hensive background of this swarm from a geographic geophysical, and tectoni c perspective. The seismicity of the area is continuously monitored by the AVO and the AEIC. The Mount Spurr seismic network which is operated by the AVO, 30 km to the southwest of Strandline Lake, is the main source of seismic monitoring in the region. However, the AEIC operates numerous seismic stations through out the state of Alaska and one of their stations (SKN) is located ~ 50 km northeast of Strandline Lake A second AEIC station (SSN) is located ~60 km east of Strandline Lake. Strandline Lake is located in a complex tectonic setting at the northern end of the Aleutian volcanic arc. As previously stated, it is located near the junction of the North America n and Pacific plates and the subducting Yakutat block (Figure 1.2) and sits between two major active fault systems, the Castle Mountain and Denali. Dextral transpression of the entire Cook Inlet area appears to be driven by coupling between the North American and Pacific plates, and by lateral escape of the Yakutat bloc k (Haeussler, 2000). A previous study of seismic activity in the Mount Spurr area (Roman et al., 200 4) and a study looking at elevated seismic activity along the entire Aleutian Arc in 1996 (McNutt and Marzocchi, 2004) both document the swarm seismicity in the Strandline Lake area While McNutt and Marzocchi, (2004) suggest that the seismic swarm at

PAGE 22

13 Strandline Lake is representative of a strain transient, it is also possible that this swarm may be a precursor to the bir th of a new volcano, due to its location along the volcanic front (Figure 1.1). Aside from this, little literature currently exists on the swarm and is limited mostly to informal communication s from the AVO. A review of the studies which reference the sw arm at Strandline Lake will be included at the end of this section.

PAGE 23

14 2.2.1 Physiographic and Geologic Setting Located a t 61.47N and 152.03 W, Strandline Lake sits within a remote re gion of the Alaskan wilderness, northwest of the Cook Inlet. It lies approximately 30 km northeast of Mount Spurr volc ano and 110 km west of the city of Anchorage (Figure 1.1) Strandline Lake sits at the foot of the Tor drillo Mountains to the west, which give way to the Susitna lowlands to the east. Strandline Lake itsel f is the result of the damming of the Beluga River by the Triumvirate Glacier. The south end of the lake is dammed by the g lacier and is subject to frequent outburst floods called jkulhlaups during melting or collapse of the glacier. However, there are n o records of a jkulhlaup in 1996 at Strandline Lake (Matthew Sturm, personal c ommunication) To make interpretations of rock units surrounding the Strandline Lake area, two geologic maps of the Cook Inlet region (Magoon et al., 1976 and Wilson et al., 2009) were reviewed. The lake is bounded to the east and north by L ate Cretaceous/ E arly Jurassic low grade metamorphosed rock, consisting primarily of slate and rocks of volcanic origin (Figure 2.3). To the west, th e lake is bounded by Paleocene plutonic rocks composed mostly of granite, quartz monzonite, and syenite The area around Strandline Lake also contains rocks of volca nic origins To the north and east of Strandline Lake several Paleocene to E arly Cretaceous volcanic intrusions are present. To the north and southwest o f the lake, several Cretaceous volcanic intrusions are present. No Quaternary volcanic deposits, such as those found near Mount Spurr are found adjacent to Strandline Lake and no historical eruptive events have been recorded there.

PAGE 24

15 Figure 2.3 Geologic map of the Strandline Lake, Alaska area. Strandline Lake is denoted by the black box. Strandline Lake is bounded to the east by a metamorphic rock unit (Kivs), to the west by a granitic intrusive unit (Tpgr), and to the south by the Triumvirate glacier. Volcanic intrusions (TKv and Kv) also dot the landscape around Strandline Lake. Quaternary volcanic deposits (Qv), which are prominent around Mount Spurr, are absent in the area immediately surrounding Strandline Lake. For simplicity, only those units which are relevant to the region immediately surrounding Strandl ine Lake are listed above. For a full list of map units please refer to Wilson et al. (2009). (Figure modified from Wilson et al., 2009)

PAGE 25

16 2.2.2 Tectonic Setting Strandline Lake is located in a compressive tectonic regime where the oceanic crust of the Pacific plate is being subducted beneath the continental margin of the North American plat e This subduction has given rise to the Aleutian volcanic arc, one of the most active volcanic arcs in the world. It has also been the cause of some of the most intense earthquakes ever recorded in human history (Plafker, 1965; Kanamori, 1971) Additionally, the Yakutat block, an exotic terrane, is actively accreting and subducting along the southern margin of Alaska (Brocher et al., 1994) It is widely thought that the three tectonic plates/blocks converge somewhere near Strandl ine Lake (Figure 1.2 ). Strandline Lake is directly aligned with the volcanic front formed by the active volcanoes in the Cook Inlet region (Mount Spurr, Redoubt, Iliamna, Augustine, and F ourpeaked) (Figure 1.2 ) Thus, Strandline Lake sits in the correct location to be a n incipient ( crypto ) volcano. Additionally, a s seen in F igure 2.3, small un named faults striking in a general NW SE direction have been mapped in the area around Strandline Lake. The area sits at the boundary between the Aleutian megathrust and the Alaska Range between the Denali and Lake Clark Castle Mountain fault systems (Figure 1.2 ) The area is therefore subjected to a wide range of tectonic activity and the observed seismic activity may be a result of stress transfer from the larger fault systems to the smaller un name d faults. However, this hypothesis becomes problematic due to the fact that a main shock aftershock sequence, commonly associat ed wit h brittle failure of rock under pure tectonic stresses, has not been observed An alternative hypothesis is that fluid upwellings from the edge of the subducting slabs play a key part in the seismogenesis of the swarm

PAGE 26

17 2.2.3 Geophysical Monitoring The seismicity of the Strandline Lake area has been monitored continuously since Seismic monitoring is the only method of continuous geophysical monitoring currently being conducted at Strandline Lake. Data from g as monitoring overflights and a global positioning system (GPS) network do not exist for this area. The bulk of the seismic network consists of seismometers installed and operated by the AVO on Mount S purr and Crater Peak (Figure 2.4 ). However, for larger S trandline Lake earthquakes, arrivals recorded by the seismic network of Redoubt volcano 125 km to the south, are useable for event analysis as well. T hree seismometers (AVO station STLK and AEIC station s SKN and SSN ) to the east and north of Strandline L ake are vital to this study due to the fact that they provide azmiuthal coverage for determining precise locations and fault plane solutions for the events. Station SKN, north of Strandline Lake, has been in operation since 1991 and recorded a large numbe r of the events from the 1996 1998 swarm. Station STLK, just east of Strandline Lake, was installed by AVO in 1997 midway during the swarm, and thus only the second half of the swarm was recorded on this station. Station SSN, to the east of Strandline Lake is an AEIC station located on Mount Susitna which recorded many of the 1996 1998 swarm events.

PAGE 27

18 Figure 2.4 Seismic network surrounding Strandline Lake The majority of monitoring stations (red triangles) come from the Mount Spurr network to the south. Station STLK was added in 1997, during the middle of the swarm. Station s SKN, to the northeast, and SSN, to the east, close the azmiuthal gap when constraining earthquake locations. 15 2 o 152 o 15 1 o 3 0 15 1 o 00 15 0 o

PAGE 28

19 2.2.4 The 1996 1998 Earthquake Swarm In late August 1996 a n intense seismic swarm began beneath Strandline Lake, Alaska. The swarm formed a dense cluster initially thoug ht to be centered between 5 and 7 km BSL (Figure 2. 5 ) The AVO noted this seismicity in their day t o day monitoring of Mount Spurr; however, Strandline Lake events were seen as anomalous and classified separ ate ly The AVO catalog indicates a total of 2,999 events recorded in the immediate vicinity of Strandline Lake during the two year period. Figure 2. 6 displays a histogram of th e seismic activity at Strandline Lake before, during, and after the swarm. A ctivity reached its peak intensity during September 1996, when 42 events were recorded during one 24 hour period Swarm earthquakes typically have an average magnitude of less than M w 1.0. T he largest earthquake recorded during the Strandline Lake swarm w as a M w 3.1 on July 1, 1997, almost exactly at the mid point of what is defined as the swarm period (Figure 2.7 ). Intense seismic activity continued until late August, 1998. The cumulative seismic moment released during the swarm is 2.03e 15 N m, or the e quivalent of a M w 4.2 earthquake (Figure 2.8) In the period following the swarm, seismic activity was not as energetic, however, as F igure 2. 6 shows, it remained noticeably higher than pre swarm conditions. This may be due to the presence of a local sei smic station (STLK) that did not exist prior to the swarm. It is not entirely clear whether the swarm was caused by t ectonic or magmatic processes. H owever, tectonic earthquakes typically show a pattern of a large earthquake (mainshock) followed by numerous smaller earthquakes (aftershocks) where as volcanic

PAGE 29

20 Figure 2.5 Plot showing the catalog depth s of the swarm earthquakes over time. The se depths were determined prior to development of the 1 D velocity model for Strandline Lake. Figure 2. 6 Histogram of the seismic activity at Strandline Lake, Alaska over a 10 year period. The onset of the swarm in question is clearly visible in the middle of 1996 After the end of the swarm in mid 1998, seismic activity remained elevated beyond the levels of pre swarm activity, but relatively quiet compare d to the swarm itself. (Figure from Roman et al., 2004) 40 35 30 25 20 15 10 5 0 5 07/04/96 11/06/96 03/11/97 07/14/97 11/16/97 03/21/98 07/24/98 11/26/98 Depth (km) Depth vs. Time

PAGE 30

21 Figure 2.7 Plot of earthquake magnitudes over time for earthquakes at Strandline Lake. The onset of the swarm is denoted by the red arrow. The largest events ( > M 2.0) are clearly seen to occur during the middle of the swarm. Figure 2. 8 Plot showing the cumulative release of seismic energy from the earthquake swarm at Strandline Lake. The onset of the swarm is denoted by the red arrow.

PAGE 31

22 ones do not. Yet, the lack of gas, thermal, and continuous deformation monitoring means that few ot her signs of possible volcanic unrest can be diagnosed. The European Remote Sensing (ERS) 2 satellite made several orbits (before, during, and after the swarm) during which SAR images of Strandline Lake were acquired, allowing for analysis of surface defo rmation during the swarm in addition to seismological analysis of swarm earthquakes.

PAGE 32

23 2.2.4.1 Previous Work Several published studies have referenced seismicity at Strandline Lake (Jolly and Page 1994; Flores and Doser, 2005) and one study (McNutt and Marzocchi, 2004) has attempted to e xplain th e cause of the swarm Jolly and Page (1994) described seismicity at Strandline Lake two years prior to the onset of the swarm in an analys is of seismicity around Mount Spurr. Earthquake depths from the surface to 20 km BSL were reported at Strandline Lake during this time. Jolly and Page (1994) state that earthquake depths observed at Strandline Lake are consistent with maximum crustal dep ths observed elsewhere in south central Alaska. Flores and Doser (2005) reference seismicity at Strandline Lake in their analysis of shallow seismicity in the Anchorage, Alaska area between 1964 and 1999 They describe seismicity at Strandline L ake as a shallow (<10 km) cluster which appears to show a tabula r feature at 9 10 km depth. The y further show that in cross section, seism ic activity at Strandline Lake appears to be spatially independent of the Castle Mountain fault syst em. Volcanic activity is presented as a possible cause of the seismicity, but this hypothesis A statistical approach was used by McNutt and Marzocchi (2004) in an attempt to explain the likelihood of multiple volcanic eruptions and seismic swarms happening sim ultaneously in Alaska during the fall of 1996. They describe the Strandline Lake swarm as deep (3 12 km) and even though no volcano exists at the site, they include it in their analysis of volcanic swarms under the assumption that because it lies within a volcanic arc, the seismicity at Strandline Lake is related to the same stresses that result in

PAGE 33

24 the formation of volcanic centers. Using relationships between swarms and eruptions at Iliamna, Martin Mageik, Strandline Lake, and Peulik they speculated tha t the most likely mechanism for arc wide unrest in 1996 was a deformation pulse or strain transient which affected the eastern half of the Aleutian arc.

PAGE 34

25 Chapter 3 Methods The purpose of this research is to investigate the earthquake swarm at Strandline Lake, Alaska and to determine its driving mechanism. To do so, five main methods are used: (1) Calculation and analysis swarm and background statistics (2) d evelop ment of a new 1 D velocity model approp riate for the subsurface at Strandline Lake, ( 3) relocation of Strandline Lake earthquake s from the AVO catalog using the new velocity model, ( 4) c alculation and analysis high quality fault plane solutions, and (5 ) deformation analysis of InSAR images. Wi thin this chapter a detailed description of the procedures and programs used in this research is provided.

PAGE 35

26 3.1 B values In reviewing the literature on non volcanic earthquake swarms, one common theme is apparent; b values are often used to augment other structural data, deformation measurements made by global positioning satellites (GPS), and monitoring of gas emissions ( e.g., Brauer et al, 2009; von Seggern et al., 2008; Fischer and Horalek, 2005; Smith et al, 2004). B values indicate the magnitud e distribution of the earthquakes within a swarm and are determined from the slope of the frequency magnitude regression curve in a log linear plot. Swarms and aftershock sequences driven by regional tectonic activity typically have a b le swarms driven by intruding fluids and/or magma can have b values greater than 2.0 (Smith et al., 2004; Jakobsdottir et al., 2008). Analysis of b values combined with analysis of the earthquakes, available deformation and gas data, and observations of t he overall temporal and spatial evolution of the swarm may provide information about the source mechanism of the swarm. To calculate the b values for the background and swarm periods for the Strandline Lake area events from the AVO earthquake catalog are first sorted by magnitude. The total number of earthquakes for each magnitude from M w 0.9 to 3.7 by increment of 0.10 is then cumulatively tallied. The log of each of these tallies is then calculated and plotted on the y axis along with the correspondin g magnitude values on the x axis. A regression line is added to the plot, but only takes into account magnitude values above the threshold of completeness (the point at which < 100% of the earthquakes for a given increment are reported due to equipment de tection limits) The slope of this regression line corresponds to the b value for the given events Therefore, data sets with a large fraction of smaller earthquakes should have a steeper slope and yield a higher b value and vice versa.

PAGE 36

27 3.2 Velocity Mo del An important part of this research is the development of a new 1 D velocity model. Deve lopment of this model is completed using the program VELEST and i s critical to the subsequent relocation of Strandline Lake earthquakes because locations of earthquakes are inherently dependent on the velocity model used. The AVO locates events in the Strandline Lake area using the 1 D velocity model developed by Jolly and Page (1994) for Mount Spurr (Table 3.1) However, horizontal and vertica l location errors, along with root mean square (RMS) residuals for events located using the Mount Spurr velocity model are high and do not provide a high degree of confidence. Earthquakes are located on a second run with an additional velocity model, the generic Alaska velocity model (Table 3.2) to provide an added degree of confidence and another data set for comparison during analysis of the results. Diagrams of both the Mount Spurr and Generic Alaska velocity models are shown in Figure 3.1. The program VELEST (Kissling, 1995) inverts phase arrival times for a s et of input earthquakes for the best 1 D velocity model (V p and V s ) along with individual station corrections. The first step in thi s process is to acquire a set of repicked Strandline Lake earthquakes which have clear and impulsive first arrivals. This i s done by going through the waveforms from each station f or each earthquake within the AVO catalog and placing a pick only on impulsive P and S wave arrivals (Figure 3. 2 ). In this way error associated with variation in user picks is minimized. Those earthquakes with enough clear first arrivals are then located with the program Seisan (Havskov and Ottemoller, 200 0 ) using the Mount Spurr velocity model (Table 3.1) It is important to note that to maintain an azmiuthal gap of less than 180 degrees, all earthquakes within

PAGE 37

28 Mount Spurr Velocity Model Depth to Top of Layer BSL (km) V p (km/s) V s (km/s) 3.0 5.10 2.82 2.0 5.50 3.04 5.3 6.30 3.62 27.3 7.20 4.04 Table 3.1 The four layer Mount Spurr 1 D velocity model developed by Jolly and Page (1994). The AVO currently uses this model to locate events in the Strandline Lake area. Generic Alaska Velocity Model Depth to Top of Layer BSL (km) V p (km/s) V s (km/s) 3.0 5.30 2.98 4.0 5.60 3.15 10.0 6.20 3.48 15.0 6.90 3.88 20.0 7.40 4.16 25.0 7.70 4.33 33.0 7.90 4.44 47.0 8.10 4.55 65.0 8.30 4.66 Table 3.2 The nine layer generic Alaska 1 D velocity model. This model was used in addition to the Mount Spurr velocity model to provide an extra data set f or analysis

PAGE 38

29 Figure 3.1 Diagram of the Mount Spurr and Generic Alaska velocity models. (a) The Mount Spurr velocity does not provide much structure to the upper 20 km of the crust. (b) The Generic Alaska velocity model appears more robust, earthquakes located with it in the Strandline Lake area have high horizontal and vertical errors 30 25 20 15 10 5 0 4.00 4.50 5.00 5.50 6.00 6.50 7.00 7.50 Depth BSL (km) V p (km/s) Mount Spurr Velocity Model 70 60 50 40 30 20 10 0 4.00 5.00 6.00 7.00 8.00 9.00 Depth BSL (km) V p (km/s) Generic Alaska Velocity Model (a) (b )

PAGE 39

30 Figure 3 2 Example earthquake seismograms from Strandline Lake. Top: Impulsive P wave from a Strandline Lake earthquake showing a downward first motion. User error in making a pick is minimized when using high quality waveforms such as this. Bottom: E mergent waveform from a Strandline Lake earthquake. A high quality pick on the first arrival of this earthquake is not possible. Such earthquakes are not used in developing the 1 D velocity model. this set must have a high quality first arrival pick on station SKN. The phase arrival picks, time, and location of each of the high quality e arthquakes are then input into VELEST along with the name s and location s of the surrounding AVO and AEIC sta tions and the number of velocity model layers to be solved for An initial velocity model is input as a starting reference for the program. For this study a modified version of the generic Alaska velocity model is used as the reference model because fiv e of the nine layers in the generic Alaska velocity model are within the first 20 km BSL as opposed to only three layers in the Mount Spurr model. The remaining four layers of the generic Alaska model were omitted due to the fact that the majority of the Strandline Lake earthquakes appeared to occur with in the upper 20 km of the crust based on P and S wave separation times Using an iterative process, VELEST then inverts for the 1 D velocity model which results in the lowest average R MS value for the input earthquakes If there is a large (>1.0 0 km/s) difference in V p between adjacent layers in the resulting

PAGE 40

31 model an additional layer is inserted and the process is repeated. If the difference in V p is small (<0.15 km/s), a layer is deleted and the process is repeated until a stable solution resulting in the lowest average RMS is found

PAGE 41

32 3.3 Hypocenter Relocations Relocation of the events is done with the program HYPOELLIPSE and the input of the new 1 D velocity model. The e arthquakes at Strandline Lake are currently located (in the AVO catalog) using the Mount Spurr velocity model developed by Jolly et al (1994). The depths of the swarm earthquakes in the AVO catalog are between 3 12 km BSL (McNutt and Marzocch i, 2004). Using the program HYPOELLIPSE (Lahr, 1999) all swarm events are relocated and examine d here to determine precise 3 dimensional locations of the earthquakes HYPOELLIPSE is a suitable program to use because of its ability to account for large topographic variations. This is accomplished b y embedding the seismic station s within the newly developed Strandline Lake veloci ty model, which extends 3 km above sea level (ASL) Three trials are conducted in this study ; one for each of the Mount Spurr, Generic Alaska, and new Strandline Lake velocity models Arrival times from the AVO catalog are input into HYPOELLIPSE along with a velocity model and its associated station corrections. Travel times are then Method to minimize the RMS of the travel time residuals (L a h r, 1999). The equation used t o determine the RMS residual is = 1 2 1 1 2 ( 1 ) where R i is the observed minus the computed time for the i th phase, and W i is the user specified weight of the i th phase Each of the output data sets are then plotted using Generic Mapping Tool s (GMT) in map view to highlight lateral differences in hypocenter

PAGE 42

33 locations as well as in cross section to show shifts in hypocenter depths. Horizontal and vertical location errors, along with RMS residuals for relocated earthquakes are then examined an d compared to determine the effect of relocation with different velocity models.

PAGE 43

34 3.4 Fault Plane Solution Analysis By analyzing the fault plane solutions of Strandline Lake earthquakes, the crustal stress field during the swarm can be under stood. Strandli ne Lake may lie within a transitional tectonic regime due to interaction between the three converging tectonic blocks and understanding the crustal stress es beneath it may be key to understanding the cause of the 1996 1998 earthquake swarm. The a pproach taken in this study is to attempt to calculate fault plane solutions only for the largest magnitude Strandline Lake earthquakes, which in this case is the same set used in developing the 1 D velocity model. F ocal mechanisms are calculated using th e program Seisan (Havskov and Ottemoller, 2001) and FPFIT ( Reasenberg and Oppenheimer, 1985 ) Seisan is a set of programs used to analyze and locate earthquake s. FPFIT calculates a fault plane solution based on the first motion polarities and locations of the earthquake ; henc e the need for high quality events when initially searching through the data set. The FPFIT calculation is accomplished through a two stage grid search procedure that finds the source model minimizing a normalized, weighted sum of f irst motion polarity discrepancies. Two weighting factors are incorporated in the minimization: one reflecting the estimated variance of the data, and one based on the absolute value of the theoret ical P wave radiation amplitude. In addition to finding th e minimum misfit solution, FPFIT finds alternative solutions corresponding to significant relative minima in misfit. Such solutions, when they exist, generally correspond to faulting mechanisms distinctly different from the minimum misfit solution, and may be deemed the preferred solution after consideration of possible data errors, unmodeled

PAGE 44

35 refractions and a priori knowle dge of the tectonic environment (Reasenberg and Oppenheimer, 1985). After FPFIT outputs the set of candidate fault plane solutions, the y are manually examined to determine whether they meet additional quality criteria The conditions each solution must meet to be considered for stress analysis are as follows: (1) the RMS for the earthquake location must be below 0.20 seconds, (2) the azm iuthal gap must be less than 180 degrees, (3) the horizontal and vertical error must be less than 5 km each, (4) the percentage of misfit polarities or first motion s plotted in a quadrant of the fault plane solution which represents the opposite polarity must be less than 10% and (5) the strike, dip, and rake uncertainty must be less than 20 If multiple solutions exist for an earthquake that meet s the above criteria the earthquake is removed from further analysis on the basis of ambiguity in the data Only the high qu ality fault plane solutions ar e given and analyzed in the following chapter s Additional analysis of the crustal stress field is done using plots of R ose diagrams and stereonets For this study, the orientations of p 1 ) and t 3 ) for all high quality earthquakes are plotted in a Rose diagram along with their corresponding stereonets to highlight the main trends Data used in the plots are taken from the orientations of the nodal planes of each fault plane solution s cal culated by FPFIT All plotting is done using custom GMT scripts.

PAGE 45

36 3.5 Surface Deformation For this study, InSAR images were used to determine whether ground deformation occurred during the 1996 1998 earthquake swarm at Strandline Lake, Alaska. Most impor tantly, InSAR imagery may provide insight into the roles played by magma or geothermal fluids during this swarm which seismic data analysis is unable to do and may aid in pinpointing a likel y source mechanism A GPS monitoring network d oes not exist in the remote regions surrounding the Cook Inlet in Alaska. In an attempt to supplement the seismic data, deformation analysis is accomplished b y analyzing the InSAR images Image pairs from before, during, and after the swarm period are an alyzed to detect any temporal changes in ground elevation. A schematic diagram showing the process in which SAR i mages are collected and the principle of SAR imagery is shown in Figure 3. 3 Images used in this study were taken by the SAR instrument aboar d the ERS 2 satellite. The ERS 2 satellite was launched in April 1995 and frequently made passes over the southern Alaska region A search of ERS 2's archives yielded 49 images along descending track 229 which captured the Strandline Lake area Of the 49 images, 18 were selected based upon time of capture (1995 1999) relative to the swarm Four of the selected images were taken prior to the swarm (1995 1996), six during the swarm (1996 1998), and the remaining eight were taken post swarm (199 8 1999). Interferograms were then processed using the Repeat Orbit Imagery Package (roi_pac) software (Cal Tech/JPL) at the RSMAS, University of Miami. For successful interferogram processing, the two images to be compared needed to have a small

PAGE 46

37 perpendi cular baseline and to cover a short time span. Limits of 250 m eters and three years were chosen as the boundaries of these parameters and a total of 56 interferograms (event pairs) met these conditions. Of the se 56 event pairs only those with highly coh erent signals were used in further analysis. This reduced the final number of event pairs to 18. Topographic effects in the interferograms were removed during the interface with a National Elevation Dataset ( NED ) 2 arc second Digital Elevation Map ( DEM ) obtained from the U.S. Geological Survey National Map Seamless Server. Each color cycle (fringe) on the interferogram represents 2.8 cm of deformation, and error associated with atmospheric properties is typically < 1 cm. Figure 3.3 Illustration showing the principles of SAR imagery. An image is collected from a satellite scan during its orbit. After an amount of time has passed, a second image is taken from the same satellite on the same pass. The two images are combined to highlig from www.geodesy.miami.edu/sar.html )

PAGE 47

38 Chapter 4 Results In this chapter, the resul ts of the analysis of the 1996 1998 swarm at Strandline Lake are presented. The data presented herein highlight the main line s of support behind the conclusions drawn in subsequent chapters. This is achieved by g raphical representation s of the b values, the Strandline Lake velocity model, the 3 D relocation of the AVO catalog using the Strandline Lake velocity model background and swarm stress characteristics using fault plane solutions and stereonets and the images acquired from the InSAR processing. With these results I hope to quantify similar and contrasting characteristics of the swarm using the three aforementioned velocity models. With the exception of the locations of the earthquakes within the AVO catalog using the Mount Spurr velocity model all results are my own work.

PAGE 48

3 9 4.1 B values B values for the background and swarm periods were calculated and the frequency magnitude distributions of the data with corresponding best fit lines are presented in Figure 4.1. The graphs illustrat e a lin ear function made by the data points in log linear space the slope of which is the sought after b value. A slope of 2.0 or greater implies that a relatively large amount of the seismic energy was released by smaller magnitude events, and such a slope is characteristic of seismic sequences in volcanic areas (McNutt, 2002) although there are multiple examples of VT swarms with b values ~1.0 Conversely, seismic sequences driven by tectonic activity tend to yield a b value near 1.0 (Frohlich, 1993) Near S trandline Lake, a total of 11,138 earthquakes were recorded between January 1, 1995 and August 31, 1996 and between September 1, 1998 and September 12, 2008 These earthquakes constitute the data used for the calculations of the background b value at Str andline Lake. From Figure 4.1a, the slope of the best fit line is 1.25 (corresponding to a b value of 1.25) Note that the best fit line only accounts for earthquake magnitudes above the threshold of completeness, which for the backgrou nd period is M L 0 .3. Earthquakes with magnitudes below this threshold are not incorporated into the final b value calculation For the 2,999 swarm events which occurred between September 1, 1996 and August 31, 1998, the best fit line in Figure 4.1b yields a b value of 1 .33 (Figure 4.1b) The threshold of completeness for the swarm period is M w 0.6, slightly higher than that of the background period.

PAGE 49

40 Figure 4.1 Strandline Lake b values. (a) B value of background earthquakes in the Strandline Lake area The red line denotes the best fit line for the data up to the threshold of completeness. (b) B value of Strandline Lake earthquakes in the AVO catalog which occur between September 1, 1996 and August 31, 1998 (the swarm period) (a) ( b )

PAGE 50

41 4.2 1 D Velocity Model The development of a 1 D velocity model specific to the Strandline Lake area is an integral component of this study because it plays a key role in the new earthquake locations After several tr ia ls u sing the program VELEST, a new se v en layer 1 D velocity model was developed for Strandline Lake and is presented in Table 4.1 and a diagram of it is shown in Figure 4.2 The seven layers span a depth of 3 km ASL to 25 km BSL. Of the seven layers, six of them occur in the upper 20 km of t he crust, giving a more detailed velocity structure for the upper crust of the region than the other two models used by the AVO and AEIC. VELEST does not output values for S wave velocities, therefore the average V p /V s ratio of the Generic Alaska velocity model of 1.78 is a ssumed in the Strandline Lake velocity model and the V s values obtained using this ratio are presented in Table 4.1. Seismic velocities in the upper crust at Str andline Lake are slower than those given by the Mount Spurr and Generic Alaska velocity models. When the Strandline Lake velocity model is used to relocate Strandline Lake earthquakes the resulting RMS and location errors are dramatically reduced (Figures 4. 3 4. 4 and 4. 5 ). Locations and l atitudinal and longitudinal location errors ( represented by 95% confidence error ellipses for the Mount Spurr, Generic Alaska, and Strandline Lake velocity models ) for the 28 large earthquakes used in developing the new velocity model are shown in Figures 4. 3 4. 4 and 4. 5 The Generic Alaska velocity model produces the largest location error s ( Figure 4. 3 ). The Mount Spurr velocity model noticeably reduces the size of the error ellipses, however, error values are still relatively high (Figure 4.4 ). Errors associa ted with the location of the 28 earthq uake s using the new Strandline Lake velocity model are greatly reduced. The Strandline Lake velocity model also r educes the

PAGE 51

42 S trandline Lake Velocity Model Depth to top of layer (km) V p (km/s) V s (km/s) 3.0 4.89 2.74 1.0 5. 21 2.92 2.0 5 70 3.20 4.0 6. 04 3.39 9.0 6.47 3.64 15.0 6 63 3.72 25.0 6.81 3.82 Table 4.1 New seven layer 1 D velocity model for the Strandline Lake area. P and S wave velocities are shown from 3 km ASL to a depth of 33 km BSL. Figure 4.2 Diagram of the Strandline Lake velocity model. The new model provides more structure to the upper 20 km of the crust beneath Strandline Lake than the Mount Spurr velocity model and reduces the high horizontal and vertical errors associated with the Gene ric Alaska velocity model. 27 22 17 12 7 2 3 4.00 4.50 5.00 5.50 6.00 6.50 7.00 Depth BSL (km) V p (km/s) Strandline Lake Velocity Model

PAGE 52

43 average RMS from 0.16 to 0.11s, and the average horizontal and vertical location error from 3.3 to 2.5 km and 4.7 to 3.0 km, respectively in comparison to the Generic Alaska velocity model Depths of the 28 earthquakes determined using each velocity model show vertical shifts in the swarm locati on as well as a reduction in vertical error calculations. A north south cross section showing the 28 Strandline Lake events is presented in Figure 4. 6 Locations of the earthquakes using the Generic Alaska velocity model are poorly constrained with a large error estimate and often give a n unrealistic result of 3 km or less for the depth. Locations using the Mount Spurr v elocity model have a depth range between 5 and 13 km but have a scattered distribution. The Strandline Lake velocity model shows a tight clusteri ng of the earthquakes between 12 and 15 km. The St randline Lake model also results in the lowest vertical location errors of any of the three models tested.

PAGE 53

44 Station Name Correction (s) Station Name Correction (s) BGL 0.41 IVS 0.02 BGR 0.38 MSP 0.47 BKG 0.42 NCG 0.67 BRLK 0.05 NCT 0.01 CGL 0.81 PDB 0.11 CKL 0.59 PLR 0.72 CKN 0.21 PMS 0.66 CKT 0.47 PRG 0.63 CNP 0.45 PWA 1.05 CP2 0.77 RDN 0.15 CRP 0.65 RDT 0.33 CUT 1.71 RDW 0.09 DFR 0.19 RED 0.32 GHO 0.77 REF 0.14 IL1 0.47 RS2 0.38 ILI 0.15 RSO 0.11 ILS 0.06 SKN 1.37 ILW 0.04 SLK 0.24 INE 0.09 SPU 0.54 IVE 0.01 SSN 0.25 Table 4.2 Station corrections (in seco nds) for the Strandline Lake velocity model. Corrections are for both AVO and AEIC seismic monitoring stations.

PAGE 54

45 Figure 4. 3 Map showing the locations and 95% confidence error ellipses of the 28 large Strandline Lake earthqu akes obtained using the Generic Alaska velocity model.

PAGE 55

46 Figure 4. 4 Map showing the locations and 95% confidence error ellipses of the 28 large Strandline Lake earthquakes obtained using the Mount Spurr velocity model.

PAGE 56

47 Figure 4. 5 Map showing the locations and 95% confidence error ellipses of the 28 large Strandline Lake earthquakes obtained using the Strandline Lake velocity model.

PAGE 57

48 Figure 4. 6 North South cross section s of the 28 large Stran dline Lake earthquakes using (a) the Generic Alaska velocity model (green circles), (b) the Mount Spurr velocity model (red triangles) and (c) (next page) the Strandline Lake velocity model ( blue squares ). Error bars associated with the depth and latitudinal location of each earthquake are added. 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 5.00 61.2300 61.2800 61.3300 61.3800 61.4300 61.4800 61.5300 61.5800 61.6300 61.6800 Depth (km) Latitude North South Cross Section Generic Alaska Velocity Model 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 5.00 61.2300 61.2800 61.3300 61.3800 61.4300 61.4800 61.5300 61.5800 61.6300 61.6800 Depth (km) Latitude North South Cross Section Mount Spurr Velocity Model (a) ( b ) N S S N

PAGE 58

49 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 5.00 61.2300 61.2800 61.3300 61.3800 61.4300 61.4800 61.5300 61.5800 61.6300 61.6800 Depth (km) Latitude North South Cross Section Strandline Lake Velocity Model ( c ) N S

PAGE 59

50 4.3 Earthquake Locations The previous section presented results related to the locations of the 28 large Strandline Lake earthquakes used in the development of the Strandline Lake velocity model. This section is devoted to the locations of th e entire AVO catalog ( 14,138 earthquakes) of Strandline Lake events obtained using the new velocity model Using the Mount Spurr velocity model in the program HYPOELLIPSE, all Strandline Lake events within the AVO catalog are located and are presented in map view in Figure 4. 7 a. Th e earthquakes appear to be centered beneath Triumvirate glacier, just west of S trandline Lake itself. The average depth of an earthquake within this data set is 6.76 km BSL. Using the Mount Spurr velocity model, approximately 80% of the earthquakes within the AVO catalog are located at depths of less than 10 km BSL. When the Strandline Lake velocity model along with the new station corrections, is substituted in to HYPOELLIPSE and the AVO catalog is re located, the swarm appears Lake and the Triumvirate glacier (Figure 4. 7 b). However, the most noticeable change is in the new depth locations. A cross section in Figure 4. 8 shows the depths of the Strandline Lake events obtained using the new velocity model. The average depth of the re lo cated events is 12.14 km BSL and using the Strandline Lake v elocity model, approximately 20% of the earth quakes within the AVO catalog are located shallower than 10 km BSL. The deepest of these events are located near the base of the brittle crust Ther e is a shoaling of events located between 8 and 9 km BSL which appear s to form a horizontal linear feature in the cross section in Figure 4. 8 (n ote that this corresponds to a layer boundary in the velocity model and is thus an artifact)

PAGE 60

51 Figure 4. 7 (a) Epicenters of Strandline Lake earthquakes using HYPOELLIPSE and the Mount Spurr velocity model. (b) Epicenters of Strandline Lake earthquakes using HYPOELLIPSE and the Strandline Lake velocity model. (a) (b )

PAGE 61

52 Figure 4. 8 North South cross section of the relocated earthquakes using the Strandline Lake model corresponding to Figure 4. 7 b. S N

PAGE 62

53 4.4 Stress Analysis Using the methods outlined in Chapter 3, fault plane solutions (FPS) are calculated for eight large background earthquakes near Strandline Lake (Figure 4. 9 ). Seven of the eight earthquakes show predominantly strike slip mot ion Only one of the background earthquakes deviates from the rest and shows an almost pure dip slip motion Four of the eight background events are located in the immediate vicinity of Strandline Lake, west northwest of station STLK. The other four events are more distant from Strandline Lake and are located north northeast of station STLK. Rose diagrams and steronet plots of the P axis ( 1 /maximum compression) and the T axis ( 3 /minimum compression) of t he eight background earthquake FPS are shown in Figure 4. 10 The R ose diagram shows the P axis / 1 orientation having a somewhat scattered distribution; however, a predominant northeast southwest orientation is apparent The stereonet plot for the P axis of the background earthquakes also shows a scatt ered distribution with most of the P axes exhibiting a steep dip. The T axis/ 3 orientation exhibits a dominant northwest southeast orientation, perpendicular to that of 1 and is thus consistent with the observed predominance of strike slip faulting. The stereonet plot for the T axis of the background earthquakes shows a clustering in the northeast and southea st quadrants. FPS of th e swarm earthquakes (Figure 4.11 ) shows a much more localized and systematic distribution. The locations of these eart hquakes are confined to the peninsula of rock between the north eastern boundary of the Triumvirate glacier and the western

PAGE 63

54 Figure 4. 9 Map of the Strandline Lake region showing focal mechanisms of eight large earthquakes recorded during the background period.

PAGE 64

55 P Axis ( 1 ) T Axis ( 3 ) Figure 4. 10 Rose diagrams and stereonet plots of the P and T axes of the eight large background earthquakes at Strandline Lake. shore of Strandline Lake. A total of twenty earthquake FPS exhibit strike slip, reverse, and dip slip faulting Rose diagrams and stereonet plots of the P and T axes are shown in Figure 4.1 2 The d ominant orientation of the P axe s is northeast southwest while the T axes are perpendicular with a northwest southeast orientation. The stereonet plot of the P axe s shows the swarm earthquakes having steep dips and clustered within the northeast and northwest quadra nts. The stereonet of the T axe s shows a more diverse distribution of dips, indicative of a mix of strike slip and thrust faulting. This is in agreement with the diversity of FPS which are calculated for the twenty swarm events. The results of the stress analysis are compared to those of previous studies and are discussed in the next chapter.

PAGE 65

56 Figure 4.1 1 Map showing the FPS of the 20 large Strandline Lake earthquakes which occurred during the swarm period ( between September 1996 and August 1998 )

PAGE 66

57 P Axis ( 1 ) T Axis ( 3 ) Figure 4.1 2 Rose diagrams and stereonet plots of the P and T axes of the twenty large swarm earthquakes at Strandline Lake.

PAGE 67

58 4.5 InSAR Analysis A total of 13 InSAR image pairs (Table 4.2) were found to have coherence in the Strandline Lake area (all 13 InSAR images can be viewed in Appendix A) The 13 image pairs are examined individually for any indication of surface deformation associated with the Strandline Lake swarm. Particular attention is given to image pairs which have dates pre and during swarm and during and post swarm. One such InSAR image is shown in Figure 4.1 3 Only five of the 13 images yielded high quality results while t he rest exhibited abundant backgro und static which appears pixilated and provides little insight into surface deformation. Factors such as the considerable topographic relief of the adjacent Tordr i llo Mountains, the abundant snow cover in the area, and the presence of a moving glacier pos sibly attributed to the low quality of the other eight InSAR images. Color bands in the high quality InSAR images of Strandline Lake area have long distances between them and appear broad. Slight (< 2cm) changes in surface elevation may hav e occurred d, however, any deformation exhibited in the InSAR images of Strandline Lake likely fall s outside the margin of error associated with the instruments onboard the ERS 2 satellite. The Triumvirate glacier is adjacent to Strandline Lake and any changes in el evation found within its area cannot be determined to be from deformation of the ground surface or due to the mechanics of the glacier. While each InSAR image of Strandline Lake has its own unique characteristics, none show the att ern indicative of ground uplift or subsidence (e.g. Mou nt Peulik volcano, 1998 (Lu et al., 2007) and South Sister volcano, 2000 (Wicks et al., 2002) ).

PAGE 68

59 InSAR Image Pairs Date 1 Date 2 0 6/19/1996 0 7/ 0 9/1997 0 7/24/1996 0 7/14/1999 10/ 0 2/1996 0 6/ 0 4/1997 10/02/1996 07/09/1997 0 6/ 0 4/1997 0 7/29/1998 0 7/ 0 9/1997 0 8/13/1997 0 7/ 0 9/1997 0 7/29/1998 0 7/ 0 9/1997 0 9/ 0 2/1998 0 7/29/1998 0 9/ 0 2/1998 0 9/ 0 2/1998 0 6/ 0 9/1999 0 9/ 0 2/1998 0 9/22/1999 10/ 0 7/1998 0 7/14/1999 0 6/ 0 9/1999 0 7/14/1999 Table 4. 3 Image pairs of the 13 InSAR images of Strandline Lake. Date 1 shows the date o n which the ERS 2 satellite first captured an image of Strandline Lake and Date 2 shows the date of the image to which Date 1 was compared. Image pairs in bold denote the high quality InSAR images. N ote that the dates span times before, during, and after the swarm. Figure 4.1 3 Example I nSAR image of Strandline Lake, Alaska. The image pair used in this figure were captured on 06/19/1996 (pre swarm) and 07/09/1997 (during swarm). The red circle denotes the Strandline Lake area.

PAGE 69

60 Chapter 5 Discussion Strandline Lake lies within a non v olcanic area and is analogous to other non volcanic areas which have experienced seismic swarms ( e.g. Chiu et al., 1982, Fischer and Horalek, 2005, Ma and Eaton, 2009, Smith et al., 2004, Vidale and Shearer, 2006) This chapter will examine the results presented in the previous chapter and compare and contrast them to published results from studies of other non volcanic seismic swarms. The results to be compared include the depth of the swarm s the largest event within the swarm s the total number of earthquakes constituting the swarm s the b value, amount of any measurable deformation, and the suggested source mechanism. B values muc h greater than 1.0 are anomalous for upper crustal tectonic earthquakes, (Fro hlich, 1993 ) but are sometimes observed for volcanic earthquake swarms ( e.g. Smith et al., 2004). Lake Tahoe, California and Upptyppingar, Iceland experienced seismic swarms re sulting from an intrusion of magma in the shallow crust and both swarms had b values greater than 2.0 (Jakobsdottir et al., 2008, Smith et al., 2004). Conversely, earthquake swarms thought to be the result of hydrothermal fluid circulation in the upper crust at Bohemia, Germany and at Mount Hood, Oregon have b values closer to 1.0 (J ones and Malone, 2005, Klinge et al., 200 3 ). At Strandline Lake, t he b values for the background and swarm period are 1.25 and 1.33 respectively. The swarm b value of 1.33 is significantly higher than the 1.0 benchmark thought to be

PAGE 70

61 characteristic of tectonically driven events (e.g. aftershock sequences) H owever, the background value is also noticeably elevated as well making the change in b value only 0.08 Because of the elevated background b value at Strandline Lake direct comparison of the swarm b value to other swarm b values may be misleading but taken together with other observations a clearer understanding may emerge. The results of the velo city model inversion yielded a seven layer model which focused on the upper 25 kilometers of the crust beneath Strandline Lake. The Strandline Lake model pro duces a large reduction in error and RMS residuals dur ing location of earthquakes in the area. Wh en compared with both the Mount Spurr and Generic Alaska velocity models, the Strandline Lake model incorporates layers w ith generally slower P and S wave velocities This may reflect the underlying lithology of the Strandline Lake region which has prop erties much different than what is represented in the other two models. Future seismological research around the Strandline Lake area should incorporate the new model in any earthquake location analysis. An important result following from use of the Strandline Lake velocity model in locating earthquakes in the region was a downward shift in hypocentral depth for the 1996 1998 earthquake swarm with respect to previously determined catalog locations The earthquake swarm at Strandline Lake is mu ch deeper than initially thought. McNutt and Marzocchi (2004) calculated that most earthquakes within the Strandline Lake swarm occurred from 3 12 km in depth (Figure 5.1). Using the Strandline Lake velocity model, most earthquakes are located at depths from 10 17 km BSL (Figure 4. 8 ). By comp aring Figures 4.7 and 5.1, it becomes cle ar that the Strandline Lake velocity model m igrate earthquake locations deeper than previously thought. The new depths are comparable to

PAGE 71

62 Figure 5.1 Map view and cross section from McNutt and Marzocchi ( 2004 ) showing seismic activity at Strandline Lake between October 1989 and December 2001. The 1996 1998 seismic swarm is clearly shown as the cluster of circles in both diagrams. the depths of the s warms at Lake Tahoe in 2003 (Smith et al., 2004) and Upptyppingar in 2007 (Jakobsdottir, 2008), which were both caused by magmatic intrusion. Earthquake swarms at Mount Hood (Jones and Malone, 2005) Bohemia (Klinge et al., 2002) Arkansas (Chiu et al., 1984) and Matsuhsihiro, Japan (Jakobsdottir et al., 2008) were compared to the Strandline Lake earthquake swarm as well (Table 5.1). Each of the four swarms occurred at relatively shallow depths and with the exception of the Arkansas

PAGE 72

63 swa rm ( for which a source mechanism is still un determined ) each i s believ ed to have been caused by processes not directly related to magmatic activity These observations s uggest that swarms which have an origin greater than 10 15 km BSL are initiated by in teractions with magma. However, Haeussler et al. (2000) show s that seismicity in the Cook Inlet area occur s as deep as 35 km BSL and argue s that 35 km BSL could be a valid representation of the base of the seismogenic zone. If so, then the seismogenic zone at Strandline Lake is considerably deeper than the 15 18 km zone at Lake Tahoe (Smith et al, 2004) and the 8 11 km seismogenic zone beneath Iceland (Clifton et al., 2003). Therefore, if the seismogenic zone indeed extends to 35 km at Strandline Lake, the 1996 1998 earthquake swarm is certainly deep, but the characteristics of the seismogenic crust make it difficult to draw fair comparisons to other deep swarms based solely on this finding Deformation measurements from GP S networks provided an additional level of confidence in the interpretations at Lake Tahoe and Upptyppingar. While GPS measurements during the Strandline Lake earthquake swarm do not exist, the InSAR images processed at RSMAS provide some insight into the magnitude of surface deformation during the swarm Five of the thirteen InSAR images ( see Appendix A: I mages A 1/2/6/ 8 / 11) of Strandline Lake show a clear view of changes in the ground elevation between capture dates. The images reveal that little to no surface displacement occurred between pre swarm and during the swarm (Appendix A: Image 1), strictly during the swarm (Appendix A: Image 6), during the swarm and post swarm (Appendix A: Images 2 and 8), and after the swarm (Appendix A: Image 11). The ima ges lack the

PAGE 73

64 Table 5.1 Comparison of results and statistics of the 1996 1998 swarm at Strandline Lake, Alaska to six other earthquake swarms which occurred in non volcanic areas.

PAGE 74

65 depth. Instead, these images show broad, gradual changes from one color band to the next wh ich frequently appear to follow the topography of the area. E ach color band on an InSAR image represents 2.8 cm of elevation change in the surface. At Lake Tahoe 5.8 mm of deformation was observed at Lake Tahoe during the 2003 earthquake swarm (Smith et al., 2004) and 2 cm of deformation was observ ed at Upptyppingar during the 2007 swarm. Therefore, while InSAR imagery did not detect uplift of the Strandline Lake area, sub cm deformation cannot be ruled out. Stress analysis for this study revealed that the P axe s of representative b ackground earthquakes are oriented in a northeast sout hwest direction and matched those of the P axes for swarm earthquakes It is a com mon occurrence for the principa l stress axes to change orientation from background to swarm time periods when a magmatic dike intrudes into the brittle crust ( e.g. Roman and Cashman, 2006). Moreover, during the research of Wi e mer et al. (1999) the background stress characteristics of southern Alaska were examined through analysis of shear wave splitting in data from several three component seismic stations including st ation SKN located north of the Strandline Lake area Figure 5.2 show s a R ose diagram from Wiemer et al. (1999) which indicates a northeast southwest orientation of split s wavelet polarizations, a proxy for the orientation of the maximum principal stress axis However, Jolly and Page ( 1994) examined the background seismicity and stress characteristics around Mount Spurr and presented fault plane solutions for 21 large earthquakes, several of which occurred in the Strandline Lake area. Figure 5.3 shows a R ose diagram of the P axes of these 21 earthquakes A dominant northwest southeast (arc perpendicular) trend is appa rent, an approximate 90 shift from the orientation found at Strandline Lake

PAGE 75

66 during this study. This suggests one of two possibilities: (1) T he background maximum compression orientation at Strandline Lake w as northwest southeast prior to the swarm as ind icated by Jolly and Page (1994) o r (2) the orientation of maximum compressive stress orientation at the Mount Spurr area is ~ 90 different from that at and north of Strandline Lake. As shown in C hapter 2, the tectonic characteristics of this area are complex and it is unclear whether the Strandline Lake area has a background crustal stress field similar to that at Mount Spurr (characterized by trench parallel compression resulting from plate conve rgence) or to that at station SKN (characterized by trench perpendicular compression resulting from southwestward escape of the Yakutat block) Thus, it is unclear whether the local stress field during the 1996 1998 swarm can be considered thus indicative of magma intrusion) or not

PAGE 76

67 Figure 5.2 Rose diagram from Wiemer et al. (1999) showing the stress orientation at seismic station SKN from split shear wave polarizations of 69 large earthquakes. The orientation is similar to that found in Figures 4.9 and 4.11 for Strandline Lake. Figure 5.3 Rose diagram compiled from P axis orientation s given by Jolly and Page (1994) for tectonic earthquakes around Mount Spurr. Several of the earthquakes used are located in the Strandline Lake area. Primary orientation of the P ax e s is to the northwest southeast, perpendicular to P axis orientations for Strandline Lake swarm earthquakes

PAGE 77

68 Chapter 6 Conclusions Through c alculation of a new 1 D velocity model and analysis of b values, stress characteristics, and InSAR images, an in depth view of the 1996 1998 earthquake swarm at Strandline Lake Alaska has been obtained However, the analytical results obtained in this study do not provide conclusive evidence for or against a deep injection of magma This still leaves unanswered t he question of what initiated an earthquake swarm in a regio n lacking a volcano It may seem logical to l abel the however, that conclusion does not explain why the earthquakes occurred in a concentrated volume without an initiating main shock. Because the earthquake swarm at Strandline Lake occurred muc h deeper than initially thought, the argument for magmatic intrusion as the source mechanism gains favor. Generally speaking, earthquakes related to tectonic stresses alone occur at depths shallower than 10 15 km in the western United States (Ord and Hobb s, 198 9 ) and earthquakes deeper than this are uncommon outside of subduction zones. The earthquake swarms at Lake Tahoe and Upptyppingar were both located deeper than 10 15 km and both were attributed to an intrusion of magma. With the improved location of the Strandline Lake swarm, it is now apparent that it too is located in this approximate depth range However, Haeussler et al. (2000) argues that the base of the seismogenic zone in the Cook Inlet could be as deep as 35 km. To assess whether the eart hquake swarm at

PAGE 78

69 Strandline Lake occurred at the base of the seismogenic crust, a determination of the depth to the moho beneath the area is critical. In doing so, it could verify or refute the suggestion of a deep seismogenic zone by Haeussler et al. (200 0) and provide more conclusive evidence for the source mechanism of the 1996 1998 earthquake swarm at Strandline Lake. In this study, the Strandline Lake swarm is found to be comparable to the swarms at Lake Tahoe and Upptyppingar. However, an important f actor to consider concerning the se other two studies is that both were monitored by continuous GPS networks. At Strandline Lake analysis of InSAR images was the only possible method of deformation analysis No deformation was detected during the analysis of these images, however, limits on the detection threshold of InSAR measurements (2.8 cm per color band) are above the level of deformation observed during the Lake Tahoe and Upptyppingar swarms. Because surface deformation less than this threshold could have occurred undetected by InSAR analysis the lack of deformation observed in the InSAR images should not automatically rule out the possibility of a magmat ic intrusion at Strandline Lake. From the stre ss analysis it can be concluded that Strandline Lake is located in an area of transitional tectonics. The P axes of the background earthquakes at Strandline Lake show an approximate rotation of 90 with respect to the earthquake P axes near Mount Spurr, 30 km to the southwest but are parallel to split shear wavelets at station SKN, approximately 55 km to the northeast Therefore, the Strandline Lake area is of considerable importance for un derstand ing changes in stress regimes arising from Alaskan tectonics.

PAGE 79

70 This study should be viewed as the first step in understanding the earthquake swarm at Strandline Lake Further research on this swarm, and on the Strandline Lake area in general, is warranted Throughout this research it has been sho wn that deep earthquake swarms in non volcanic areas are frequently accompanied by small amounts of surface deformation and therefore it is recommended that several continuous GPS stations be installed at Strandline Lake to obtain sub centimeter accuracy data during future swarms Additionally, an in depth study of s hear wave splitting, an indicator of anisotropy and crustal stress field orientations clarification of the background stress state at Strandline Lake a nd an importan t context for interpretation of the 1996 1998 swarm FPS orientations.

PAGE 80

71 List of References injection and closely monitored induced seismicity at Paradox Valley, Colorado: Bulletin of the Seismological Society of America, v. 95, p. 664 683. Brauer, K., Kampf, H., and Strauch, G., 2009, Earthquake swarms in non volcanic regions: What fluids have to say: Geophysical Research Letters, v. 36, L17309. Brocher, T. M., Fuis, G. S., Fischer, M. A. Plafker, G., Moses, M. J. Taber, J. J., and Christensen, N. I. 1994, Mapping the megathrust beneath the northern Gulf of Alaska using wide angle seismic data : Journal of Geophysical Research v. 99 11,663 11,685. Cappa, F., Rutqvist, J., and Yamamoto, K., 2009, Modeling crustal deformation and rupture processes related to upwelling of deep CO 2 rich fluids during the 1965 1967 Matsushiro earthquake swarm in Japan: Journal of Geophysical Research, v. 114, B10304. Chiu, J., John ston, A.C., Metzger, A.G., Haar, L., and Fletcher, J., 1984, Analysis of analog and digital records of the 1982 Arkansas earthquake swarm: Bulletin of the Seismological Society of America, v. 74, p. 1721 1742. Clifton, A.E., Pagli, C., Jonsdottir, J.F., Ey thorsdottir, K, and Vogfjord, K., 2003, Surface effects of triggered fault slip on Reykjanes Peninsula, SW Iceland: Tectonophysics, v. 369, p. 145 154. Eberhart Phillips, Christensen, D.H., Brocher, T.M., Hansen, R.A., Ruppert, N.A., Haeussler, P.J., and A bers, G.A., 2006, Imaging the transition f r o m Aleutian subduction to Yakutat collision in central Alaska, with local ear thquakes and active source data: Journal of Geophysical Research, v. 111, B11303. Fischer, T., and Horalek, J., 2005, Slip generated pa tterns of swarm microearthquakes from west Bohemia/Vogtland (central Eurpoe): Evidence of their triggering mechanism: Journal of Geophysical Research, v. 110, B05S21. Flores, C., and Doser, D.I., 2005, Shallow seismicity of the Anchorage, Alaska, region (1 964 1999): Bulletin of the Seismological Society of America, v. 95, p. 1865 1879.

PAGE 81

72 Frohlich, C., and Davis, S.D., 1993, Much ado about 1.0: Journal of Geophysical Research, v. 98, p. 631 644. Haeussler, P.J., Bruhn, R.L., and Pratt, T.L., 2000, Potential s eismic hazards and tectonics of the upper Cook Inlet basin, Alaska, based on analysis of Pliocene and younger deformation: Geological Society of America Bulletin, v. 112, p. 1414 1429. Hainzl, S., and Ogata, Y., 2005, Detecting fluid signals in seismicity data through statistical earthquake modeling: Journal of Geophysical Research, v. 110, B05S07. Havskov, J., and Ottemoller, L, 2000, Seisan: the earthquake analysis software for Windows, Solaris, and Linux, users manual: Institute of Solid Earth Physics, U niversity of Bergen. Hill, D.P., 1977, A model for earthquake swarms: Journal of Geophysical Research, v. 82, p. 1347 1352. Ibs von Seht, M., Plenefisch, T., and Schmedes, E., 2006, Faulting style and stress field investigations for swarm earthquakes in NE Baveria/Germany the transition between Vogtland/NW Bohemia and the KTB site: Journal of Seismology, v. 10, p. 197 211. Jakobsdottir, S.S., Roberts, M.J., Gudmundssoon, G.B., Geirsson, H., and Slunga, R., 2008, Earthquake swarms at Upptyppingar, north ea st Iceland: A sign of magma intrusion: Studia Geophysica et Geodaetica v. 52, p. 513 528. Jolly, A.D., and Page, R.A., 1994, Seismicity and stress in the vicinity of Mount Spurr volcano, south central Alaska: Journal of Geophysical Research, v. 99, p. 15,305 15,318. Jones, J., and Malone, S.D., 2005, Mount Hood earthquake activity: Volcanic or tectonic origins: Bulletin of the Seismological Society of Ameri ca, v. 95, p. 818 832. Kanamori, H., 1971, Great earthquakes at island arcs and the lithosphere: Tectonophysics, v. 12, p. 187 198. Kissling, E., Kradolfer, U., and Maurer, H., 1995, VEL Short Introduction: Institute of geophysics and Swi ss seismological service, ETH, Zurich.

PAGE 82

73 Klinge, K, Plenefisch, T., Stammler, K, 2003, The earthquake swarm 2000 in the region Vogtland/NW Bohemia earthquake recording at German stations and temporal distribution of events: Journal of Geodynamics, v. 35, p. 83 96. Lahr, J. C., 1999, HYPOELLIPSE: a computer program for determining local earthquake hypocentral parameters, magnitude, and first motion pattern: U.S. Geol ogical Survey Open File Report 19 99 23 Lahr, J.C., Chouet, B.A., Stephens, C.D., Power, J.A., and Page, R.A., 1994, Earthquake classification, location, and error analysis in a volcanic environment: Implications for the magmatic system of the 1989 1990 eruptions at Redoubt Volcano, Alaska: Journal of Volcanology and and Geothermal Research, v 62, p. 137 151. Lu, Z., Wicks, C., Dzurisin, D., Power, J.A., Moran, S.C., and Thatcher, W., 2002, Magmatic inflation at a dormant stratovolcano: 1996 1998 activity at Mount Peulik volcano, Alaska, revealed by satellite radar interferometry : Journal of G eophysical Research, v. 107. Ma, S., and Eaton, D.W., 2009, Anatomy of a small earthquake swarm in southern Ontario, Canada: Seismological Research Letters, v. 80, p. 214 223. Magoon, L.B., Adkison, W.L., and Egberg, R.M., 1976, Map showing geology, wildcat wells, tertiary plant fossil localities, K Ar age dates, and petroleum operations, Cook Inlet area: U.S. Geological Survey Miscellaneous Geologic Investigations Map I 1019, 3 sheets. McNutt, S.R., and Marzocchi, W., 2004, Simultaneous earthquake swarms and eruption in Alaska, fall 1996: Statistical significance and inference of a large aseismic slip event: Bulletin of the Seismological Society of America, v. 94, p. 1831 1841. McNutt, S.R., 2002, Volcano seismology and monitoring for eruptions: International handbook of earthquake and engineering seismology, p. 383 406. Ord, A., and Hobbs, B.E., 1989, The strength of the continental crust, detachment zones and the development of plastic instabilities: Tectonophysics, v. 158, p. 269 289. Plafker, G., 1965, Tectonic deformation associated with the 1964 Alaska earthquake: Science, v. 148, p. 1675 1687. Reasenberg, P. A., and Oppenheimer D.H., 1985, FPFIT, FPPLOT and FPPAGE: Fortran computer programs for calculating and displaying e arthquake fault plane solutions: U.S. Geological Survey Open File Report 1985 739 Roman, D.C., and Cashman, K.V., 2006, The origin of volcano tectonic earthquake swarms : Geology, v. 35, p. 457 460.

PAGE 83

74 Roma n, D.C., 2005, Numerical models of volcanotectonic earthquake triggering on non ideally oriented faults: Geophysical Research Letters, v. 32, L02304. Roman, D.C., Moran, S.C., Power, J.A., and Cashman, K.V., 2004, Temporal and spatial variation of local st ress fields before and after the 1992 eruptions of Crater Peak vent, Mount Spurr volcano, Alaska: Bulletin of the Seismological Society of America, v. 94, p. 2366 2379. Smith, K.D., Von Seggern, D.H., Blewitt, G., Preston, L.A., Anderson, J.G., Wernicke, B .P., and Davis, J.L., 2004, Evidence for deep magma injection b eneath Lake Tahoe, Nevada California : Science, v. 305, p. 1277 1280. Triastuty, H., Iguchi, M., and Tameguri, T, 2009, Temporal change of characteristics of shallow volcano tectonic earthquakes associated with increase in volcanic activity at Kuchinoerabujima volcano, Japan: Journal of Volcanology and Geothermal Research, v. 187, p. 1 12. Ukawa, M., and Tsukahara, H., 1996, Earthquake swarms and dike intrusions off the east coast of Izu Peninsul a, central Japan: Tectonophysics, v. 253, p. 285 303. Utsu, T, 1961, A statistical study on the occurrence of aftershocks : Geophysical Magazine, v. 30, p. 521 605 Vidale, J.E., and Shearer, P.M., 2006, A survey of 71 earthquake bursts across southern Calif ornia: Exploring the role of pore fluid pressure fluctuations and aseismic slip as drivers: Journal of Geophysical Research, v. 111, B05312. Von Seggern, D.H., Smith, K.D., and Preston, L.A., 2008, Seismic spatial temporal character and effects of a deep (25 30 km) magma intrusion below north Lake Tahohe, California Nevada: Bulletin of the Seismological Society of America, v. 98, p, 1508 1526. Wicks, C.W., Dzurisin, D., Ingebritsen, S., Thatcher, W., Lu, Z., and Iverson, J., 2002, Magmatic activity beneath the quiescent Three Sisters volcanic center, central Oregon Cascade Range, USA: Geophysical Research Letters, v. 29, p. 26.1 26.4. Wiemer, S., Tytgat, G., Wyss, M., and Duenkel, U., 1999, Evidence for shear wave anisotropy in the mantle wedge beneath so uth central Alaska: Bulletin of the Seismological Society of America, v. 89, p. 1313 1322. Wilson, F.H., Hults, C.P., Schmoll, H.R., Hauessler, P.J., Schmidt, J.M., Yehle, L.A., and Labay, K.A., 2009, Preliminary geologic map of the Cook Inlet region, Alas ka: U.S. Geological Survey Open File Report 2009 1108, 2 sheets.

PAGE 84

75 Appendices

PAGE 85

76 Appendix A In SAR Images of Strandline Lake Strandline Lake area denoted by solid black square 06/19/1996 07/09/1997 07/24/1996 07/14/1999 A 1 A 2

PAGE 86

77 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 10/02/1996 06/04/1997 10/02/1996 07/09/1997 A 3 A 4

PAGE 87

78 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 06/04/1997 07/29/1998 07/09/1997 08/13/1997 A 5 A 6

PAGE 88

79 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 07/09/1997 07/29/1998 07/09/1997 09/02/1998 A 7 A 8

PAGE 89

80 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 07/29/1998 09/02/1998 09/02/1998 06/09/1999 A 9 A 1 0

PAGE 90

81 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 09/02/1998 09/22/1999 10/07/1998 07/14/1999 A 1 2 A 1 1

PAGE 91

82 Appendix A InSAR Images of Strandline Lake (Continued) Strandline Lake area denoted by solid black square 06/09/1999 07/14/1999 A 1 3