INVESTIGATIONS OF MAGMATIC END-MEMBERS: SILICIC MAGMA CHAMBERS AND MAFIC DIKES

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF GEOLOGICAL AND ENVIRONMENTAL SCIENCES AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Gwyneth Retta Hughes May 2010

© 2010 by Gwyneth Retta Hughes. All Rights Reserved. Re-distributed by Stanford University under license with the author.

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/cf090yt6229

Includes supplemental files: 1. Caldera references for Chapters 2 and 3 (Caldera_index_ref.pdf) 2. Bayes Classifier Code for Chapter 3 (bayes_classifier.zip) 3. Caldera data for Chapter 2 (Arc_caldera_data.csv) 4. Caldera data for Chapter 3 (All_caldera_data.csv)

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Gail Mahood, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

David Pollard

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Paul Segall

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

Abstract

Approximately 10% of the global population, some 550 million people, live within 100 km of an active volcano, making it imperative that the causes of magma accumulation in the crust and the factors affecting the scale of subsequent volcanic eruptions are investigated and modeled. This PhD dissertation examines magmatic processes at two scales: silicic calderas represent the large but infrequent end of the continuum, whereas mafic dikes are small in scale but common.

Many previous authors have presented hypotheses for how large silicic systems develop. In order to test and refine these hypotheses, I undertook a global compilation that empirically examines how the characteristics of 140 young silicic calderas reflect their crustal-tectonic setting. Results indicate that the size and geochemistry of silicic calderas are affected by the nature of the underlying crust, the tectonic setting, and the local stress regime. For example, large, rhyolitic calderas tend to occur in continental settings under extension. There are, however, few true

“rules,” and exceptions may prove useful in analyzing how silicic magma chambers form. Based on this compilation, I present a probabilistic method for determining the tectonic-crustal setting of a given caldera from its diameter, and eruption geochemistry. Focusing specifically on arc settings, this study demonstrates that (1) the abundance of silicic calderas in a given arc is proportional to the trench-normal convergence rate, except in arcs with back-arc spreading; and (2) silicic calderas in continental arcs tend to occur farther behind the volcanic front than do more typical iv arc volcanoes, possibly because of the abundance of pre-existing structures in continental margins.

At the opposite end of the volcanic spectrum, this dissertation examines the intrusion of two mafic dikes. The first lies beneath Mammoth Mountain, California, and was associated with a 1989 seismic swarm. Based on an inversion of leveling data constrained by relocated earthquakes, I propose that a dike 2 km long, 8 km high, with 1 m of opening was intruded at 9 km depth beneath the south side of Mammoth

Mountain. The second dike investigation focused on the eruption of Miyakejima,

Japan, in 2000, associated with more than 10,000 earthquakes, several small eruptions and progressive caldera collapse. Displacements recorded by GPS stations and pre- and syn- event seismicity were used to determine a geological explanation of the event. In the proposed model, a shallow dike propagated ~30 km away from

Miyakejima for one week, stopped propagating laterally after intersecting a pre- existing fault zone, and then continued to open and grow vertically for nearly two months.

v

Acknowledgements

I would like to thank my advisors Gail Mahood and Paul Segall for leading me into interesting branches of research and providing extensive academic support over the past five years. In addition, as committee members, David Hill and David Pollard have been invaluable as sources of feedback and encouragement.

A great deal of this thesis involved reading papers in other languages. While I was able to cope with those written in romance languages, Uwe Martens, Yo

Fukushima, and Curran Hughes helped translate papers in German, Japanese, and

Russian. Members of the Segall research group were a valuable resource of advice on programming and data analysis. The Stanford Earth Sciences library staff also helped me in obtaining a variety of data and resources. My officemates Karen Knee and Matt

Coble provided much helpful feedback and camaraderie.

A special thanks is extended to my parents Sue Cornish and Steve Hughes and all family members who have unendingly encouraged (and financially supported) my academic endeavors. Finally, I could not have come this far without the support, help, advice and encouragement of my wonderful husband, Michael Cardiff.

vi

Table of Contents

Abstract ...... iv

Acknowledgements ...... vi

Table of Contents...... vii

List of Tables...... xi

List of Figures ...... xii

1 Introduction...... 1

2 Silicic Calderas in Arc Settings: Characteristics, Distribution and

Tectonic Controls...... 7

2.1 Abstract ...... 7

2.2 Introduction...... 8

2.3 Background: Choosing the Examined Parameters ...... 10

2.4 Data and Methods ...... 13

2.4.1 Data Specific to Each Caldera...... 13

2.4.2 Data Specific to Arcs and Arc Sections ...... 16

2.4.3 The Control Group: “Normal” Arc Volcanoes ...... 17

2.4.4 Across‐arc Distribution of Silicic Calderas ...... 18

2.5 Results and Discussion...... 18

2.5.1 Caldera Attributes...... 19

2.5.2 Crustal Attributes...... 19 vii

2.5.3 Tectonic Controls on Silicic Caldera Formation...... 22

2.5.4 Distance from Volcanic Front...... 29

2.5.5 Association With Structural Features...... 31

2.6 Conclusions and Implications...... 32

2.6.1 Conclusions...... 32

2.6.2 Further Implications...... 33

2.7 Tables...... 37

2.8 Figures ...... 40

3 Tectonic settings of silicic calderas: analysis of a global

compilation...... 57

3.1 Abstract ...... 57

3.2 Introduction...... 59

3.3 Previous Work ...... 61

3.4 Data...... 63

3.4.1 Caldera‐specific Parameters...... 64

3.4.2 Regional Parameters...... 66

3.4.3 Results of Data Compilation ...... 66

3.4.4 Relationship between Diameter and Composition...... 70

3.4.5 Duration of Volcanic Activity ...... 71

3.4.6 TAS Diagrams...... 72

3.4.7 Discrimination Diagrams ...... 72

3.5 A New Bayesian Classification Scheme ...... 74

3.5.1 Results...... 75

3.6 Discussion...... 79

viii

3.7 Conclusions ...... 80

3.8 Tables...... 83

3.9 Figures ...... 86

4 Reinvestigation of the 1989 Mammoth Mountain, California

Seismic Swarm and Dike Intrusion...... 95

4.1 Abstract ...... 95

4.2 Introduction...... 96

4.3 Geological Setting and Motivations...... 97

4.4 1989 Intrusion and Earthquake Swarm – Previous Work ...... 97

4.5 Methods...... 99

4.5.1 Model ...... 99

4.5.2 Trials...... 100

4.5.3 Data ...... 101

4.5.4 Inversion ...... 102

4.6 Results ...... 103

4.7 Discussion...... 104

4.7.1 Best geometry...... 104

4.8 Conclusions ...... 106

4.9 Tables...... 108

4.10 Figures...... 109

5 Investigation of Deformation Sources for the 2000 Miyakejima,

Japan, Intrusion and Seismic Swarm...... 114

5.1 Abstract ...... 114

5.2 Introduction...... 115 ix

5.3 Previous Work ...... 118

5.4 Data and initial observations...... 121

5.4.1 Continuous GPS and Deformation...... 121

5.4.2 Seismicity...... 123

5.4.3 JMA Catalogue...... 123

5.5 Modeling GPS Deformation...... 128

5.5.1 Constraining the Model...... 130

5.5.2 Inversion Methods...... 134

5.6 Results ...... 136

5.6.1 Full Event...... 136

5.6.2 First Week...... 137

5.6.3 Evolution...... 139

5.7 Discussion...... 141

5.8 Conclusions ...... 142

5.9 Tables...... 145

5.10 Figures...... 146

6 References ...... 169

7 Appendix A...... 183

8 Appendix B...... 208

9 Appendix C...... 246

x

Tables

Table 1: Tectonic settings of examined calderas and volcanoes...... 37

Table 2: Structural settings of examined calderas...... 38

Table 3: Averaged tectonic parameters of examined arcs...... 39

Table 4: Model parameters included in the naive Bayes classifier ...... 83

Table 5: Evaluation of naïve Bayes classifier ...... 84

Table 6: Classification of test calderas and ignimbrites...... 85

Table 7: Mammoth Mountain – Inversion results...... 108

Table 8: Miyakejima - Inversion results ...... 145

xi

Figures

Figure 1: Method for measuring distance of features from the volcanic front...... 40

Figure 2: Locations of examined arcs and calderas...... 41

Figure 3: Summary of silicic caldera diameters and compositions...... 42

Figure 4: Relationship between CFE composition and the nature of the underlying

crust ...... 43

Figure 5: Relationship between caldera diameter and the nature of the underlying crust

...... 44

Figure 6: Relationship between structural setting and CFE composition and caldera

diameter...... 45

Figure 7: Correlation between caldera density and other arc volcano density and

relationship between trench-normal convergence rate and caldera density,

compared to that for all other arc volcanoes ...... 46

Figure 8: Histograms of trench-normal convergence rates for 100-km arc segments

containing at least one silicic caldera, and arc segments not containing calderas47

Figure 9: Averaged caldera composition versus trench-normal convergence rate for

each arc shows no obvious relationship ...... 48

Figure 10: Histograms of subduction obliquity for 100-km arc segments containing at

least one silicic caldera, and arc segments not containing calderas ...... 49

Figure 11: Relationship between the longevity of volcanic arcs and their respective

caldera densities and volcano densities...... 50 xii

Figure 12: Relationship between the longevity of volcanic arcs and their averaged

caldera compositions...... 51

Figure 13: Caldera density as a function of arc type and subduction parameters...... 52

Figure 14: Comparison of subduction parameters for 100-km volcanic arc segments

containing calderas with all arc segments ...... 53

Figure 15: Distributions of silicic calderas and other large volcanic features behind the

volcanic front ...... 54

Figure 16: Plot of caldera distances from the volcanic front versus slab dip at 100-km

depth. There appears to be no correlation...... 55

Figure 17: Role of pre-existing structures in silicic caldera formation...... 56

Figure 18: Global distribution of the 140 examined silicic calderas, coded by size,

caldera-forming eruption (CFE) composition, and tectonic setting...... 86

Figure 19: Distributions of average diameters and CFE compositions for all examined

calderas...... 87

Figure 20: Distributions of caldera diameters differentiated by crustal and tectonic

attributes...... 88

Figure 21: Distributions of caldera compositions, differentiated by crustal and tectonic

setting ...... 89

Figure 22: Highest reported silica content for each examined caldera-forming eruption

versus average caldera diameter...... 90

Figure 23: Duration of volcanism prior to the first CFE versus highest silica content of

the CFE and caldera diameter ...... 91

Figure 24: Total alkalis versus silica diagram...... 92 xiii

Figure 25: Rb versus (Nb + Y) tectonic discrimination diagrams of Pearce et al. (1984)

for calderas in volcanic arc settings and intracontinental or oceanic hotspot

settings...... 93

Figure 26: Locations of test calderas (not included in the database) on the modified

TAS and Pearce et al. (1984) discrimination diagrams...... 94

Figure 27: Location of study with respect to Long Valley caldera, CA ...... 109

Figure 28: Close up of Mammoth Mountain area and seismicity from May 1989

through January 1990...... 110

Figure 29: Observed data and deformation model of Long Valley caldera based on

inflation of two prolate spheroids at 6 km and 12 km depth...... 111

Figure 30: Best fitting model and marginal distributions of the inversions for all trials

...... 112

Figure 31: Volume distributions of inversions for Trials 1 through 4 ...... 113

Figure 32: Location of Miyakejima and other Izu Islands ...... 146

Figure 33:JMA epicenter locations, June 26 - August 31 ...... 147

Figure 34: Change in baselines between Niijima and Kozushima before smoothing and

after smoothing...... 148

Figure 35: Locations of GEONET continuous GPS stations that were used in inversion

of deformation; JMA seismic stations used in JMA catalogue relocation; and

ERI seismic stations used to locate earthquakes from seismic swarm ...... 149

Figure 36: Annual background motion calculated from displacements between June

1997 - June 15 2000, relative to station 0241...... 150

xiv

Figure 37: Hypocenters before relocation from JMA catalogue and hypocenters after

relocation...... 151

Figure 38: Residual distances between originally relocated hypocenters and 100

relocated bootstrap samples for the first 1000 events of the swarm...... 152

Figure 39: NIED calculated focal mechanisms for JMA catalogue...... 153

Figure 40: Pre-event seismicity, 1990-2000, color-coded by magnitude ...... 154

Figure 41: ERI seismic data, relocated by Sakai et al. (2001) ...... 155

Figure 42: Horizontal deformation, Jun. 26 - Aug. 31...... 156

Figure 43: Stress modeling via Coulomb 2.0...... 157

Figure 44: Stations at which dike-normal displacements are calculated for Figures 45-

48 and 54...... 158

Figure 45: Observed and calculated displacements for the full event - June 26 - August

31, for Model F-A ...... 159

Figure 46: Observed and calculated displacements for the full event - June 26 - August

31, for Model F-B...... 160

Figure 47: Observed and calculated displacements for the full event - June 26 -

August 31, for Model F-C...... 161

Figure 48: Observed and calculated displacements for the full event - June 26 - August

31, for Model F-D ...... 162

Figure 49: Miyakejima - Conceptual representation...... 163

Figure 50: Observed and calculated displacements for the first week of the event -

June 26 – July 2, assuming no aseismic slip within the fault zone between

Niijima and Kozushima...... 164 xv

Figure 51: Observed (blue) and calculated (red) displacements for the first week of the

event - June 26 – July 2, for Model W-A...... 165

Figure 52: Observed (blue) and calculated (red) displacements for the first week of the

event - June 26 – July 2, for Model W-B ...... 166

Figure 53: Alternative models based on MCMC output from W-B ...... 167

Figure 54: Evolution models...... 168

xvi

1 Introduction

With the global population growing at the rate of 1-2% per year and consistently concentrating itself in coastal regions, where the majority of volcanoes are located, volcanic activity presents a significant hazard for humanity in the twenty- first century. About 10% of the global population, some 550 million people, live within 100 km of a volcano that has been active during historic time (Small and

Naumann, 2001). Additionally, some of the largest population centers in the world, including Tokyo, Seattle, and Mexico City, are located within areas of high potential volcanic risk (Schmincke, 2004). It is important to consider, however, that during historical time humanity has not witnessed the full spectrum of volcanic activity that geologists know to occur. For example, the largest historical eruption, in 1815 at

Tambora, Indonesia, had a volume of approximately 50 km3, about one-sixtieth the size of the relatively young eruption at Toba, Sumatra, 75,000 years ago. The Tambora eruption killed approximately 100,000 people in the surrounding area and temporarily changed the global atmosphere, resulting in, according to Lord Byron, a “year without summer” in Europe (Schmincke, 2004) and June frosts in North America

(Oppenheimer, 2003). Because population growth is particularly high in developing countries where volcanism tends to be denser and more frequent (Tilling, 2004), it is imperative that the causes of magma accumulation in the crust and the factors affecting the scale of subsequent volcanic eruptions are further investigated and modeled. In order to determine the destructive potential of future eruptions,

1 researchers must study the entire continuum of volcanic activity: from the large- volume eruption of silicic magma chambers to the subsurface movement of basalt that feeds magmatism at almost all scales.

One end-member of volcanic activity is represented by large silicic caldera- forming eruptions (CFEs) in which tens to thousands of cubic kilometers of silicic magma are catastrophically erupted from a shallow chamber over the course of a few days, spreading ash continent-wide. Three of the most dramatic examples of caldera- forming activity during the Quaternary include Long Valley, California; Toba,

Sumatra; and Yellowstone, Wyoming. A CFE volume of 300 km3 of magma or more, termed a “super-eruption” (Sparks et al., 2005), would produce enough ash and sulphur gas to cause significant global consequences. Although no CFEs have ejected magma volumes greater than a few tens of cubic-kilometers during historical time, the results of a super-eruption can be predicted from the study of volcanic products and the effects of eruptions such as that at Tambora. Potentially millions of lives would be lost in the immediate vicinity of the eruption and ash fall within a continent-wide area would result in further fatalities from particle inhalation, building collapse, loss of infrastructure, and water contamination. By comparison, smaller CFEs, such as

Tambora, occur more frequently, about once every thousand years. While it is thus not surprising that there has not been a “super-eruption” during historical time, it is possible that one will take place during the foreseeable future, and smaller CFEs will continue to punctuate human history. Hence, despite the low probability of such eruptions from large-volume magma chambers in the upper crust, the potentially

2 catastrophic effects necessitate that these types of magmatic systems be studied and understood.

At the other end of the volcanic spectrum, mafic dikes represent the fundamental way in which mantle-derived material and heat are added to the crust.

During dike emplacement, magma is intruded into the crust because the combination of pressure within a parent magma body and the local stress regime favor vertical, as well as lateral, propagation of magma. Though smaller and less destructive than explosive silicic eruptions, basaltic eruptions from vertical intrusions are relatively frequent—for example Kilauea Volcano in Hawaii has had three in the past twenty- five years—and can result in local property damage and rare fatalities. Such activity can be particularly devastating where rising magma comes into contact with surface or ground water, causing lahars or phreatomagmatic eruptions. Dike intrusions are accompanied by seismicity, crustal deformation, and volatile release such that even strictly subsurface activity can cause significant economic loss because of evacuations and decreased property values. However, these associated events can also provide information about how magma intrudes into the crust. Though the emplacement of mafic dikes and the small-scale eruption of basaltic magma do not pose a threat at the global or even regional level, their high frequency and contribution to volcanism in all tectonic settings make the continued study of dike intrusion processes essential.

Study of both mafic and silicic magmatism is not only necessitated by the hazards volcanoes present, but it is also of fundamental importance to the scientific understanding of the geological development of Earth. Throughout post-Archean geologic time, silicic magma chambers have contributed to the growth and evolution

3 of the continental crust. Upper continental crust is preserved because it has an intermediate composition making it less dense than oceanic crust and thus unable to be readily subducted. Most new crust at continental margins is constructed by a combination of island arc accretion and arc magmatism, yet island arcs and the bulk of arc magmatism are too mafic to account for the intermediate nature of upper continental crust (Christensen and Mooney, 1995; Rudnick, 1995). In order to account for this discrepancy, dacitic to rhyolitic magma bodies must form in arcs, such that they eventually evolve to have the composition of stable continental crust (Christensen and Mooney, 1995; Fliedner and Klemperer, 2000). In essence, these silicic magma bodies are the geological sutures that “stitch” accreted crust to existing continents, and stabilize continental crust by adding evolved material to the upper crust. The study of factors that contribute to large-scale silicic magmatism is therefore essential to scientific insight into how continental crust at margins changes and stabilizes through time. Because mafic intrusion is fundamental to all magmatism, including island arc and silicic magma chamber formation, it is additionally vital to understand the mechanisms by which basaltic magma is emplaced in the crust.

In my thesis I pursue four different studies in physical volcanology, each of which will contribute to the larger understanding of magma intrusion and storage in the upper crust. The first two subsequent chapters focus on the formation and global distribution of large silicic magma chambers, and the latter two chapters examine two specific examples of mafic dike intrusion.

Chapter 2 is a global examination of silicic calderas in volcanic arc settings that examines the effects of crustal setting and subduction parameters on the nature

4 and density of silicic calderas (< 2 Ma) in all major, currently active volcanic arcs.

This study seeks to explain differences in the relative frequencies of silicic calderas in different arcs and examines the spatial relationship between silicic calderas and associated volcanic fronts. Data for this chapter were compiled from a massive literature review and are presented Appendix A, with references for individual data points listed in Appendix C. This chapter was completed independently by Gwyneth

Hughes under the direction of Professor Gail Mahood. Preliminary work for this chapter was published by Hughes in Geology, 2008 with Mahood as a coauthor. The chapter in its current form was submitted to GSA Bulletin and is in revision at the time of dissertation submission.

Chapter 3 expands the database compiled in Chapter 2 to determine whether regional parameters including tectonic setting, crustal characteristics, and stress regime have systematic correlations with caldera-specific parameters such as size, composition, and geochemistry. Based on compiled data for 140 calderas, a naïve

Bayes classifier is presented that attempts to classify the regional parameters of any silicic caldera based solely on caldera-specific parameters. The efficacy of such an algorithm has important implications for determining the regional setting of ancient calderas, plutons, or even ignimbrites. Data for this chapter is presented in a data table in Appendix B with references for individual data points also listed in Appendix C.

Matlab code for the classification algorithm is located in the Supplementary Materials.

Gwyneth Hughes completed chapter 3 independently under the direction of Professor

Gail Mahood.

5

Examining volcanism at a more local scale, Chapter 4 seeks to constrain the dimensions of a dike believed to have intruded beneath Mammoth Mountain,

California, during a 1989 earthquake swarm. A leveling line north of Mammoth

Mountain recorded displacements due to dike intrusion in addition to the larger-scale inflation occurring at Long Valley during that time. Recent work by Prejean et al.

(2003) relocated earthquakes in the swarm, pinpointing the likely position and orientation of the dike. I combine constraints from the relocated hypocenters with estimates of inflation sources beneath Long Valley to re-invert the leveling data for dike dimensions. Gwyneth Hughes completed Chapter 4 independently with helpful advice from Professors Gail Mahood and David Pollard as well as Dr. John Langbein and Dr. David Hill.

Finally, Chapter 5 investigates the 2000 seismic swarm and dike intrusion in the Izu Islands of southern Japan. This study combines examination of seismic data with inverse and forward modeling of displacements recorded by GPS to study and model the long-lived basaltic dike intrusion at Miyakejima, Japan. By modeling displacements for specific time segments, constrained by observations from pre- and syn-event seismicity, this study proposes a geologically motivated explanation of the dramatic deformation. Gwyneth Hughes completed chapter 5 independently, under the direction of Professor Paul Segall.

6

2 Silicic Calderas in Arc Settings: Characteristics, Distribution and Tectonic Controls

Preliminary work for this chapter was originally published in Geology (2008), with

Gail Mahood as coauthor. This chapter has been submitted in its entirety to GSA

Bulletin again with Mahood as coauthor and was in revision at the time of dissertation submission.

2.1 Abstract

Silicic calderas represent the surface expressions of large silicic magma bodies.

Although development of sizable volumes of granite, senso lato, is considered key in stabilizing continental crust, the factors that contribute to its origin are debated. Based on a comprehensive literature review, we analyze the characteristics, tectonic settings, and spatial distributions of 108 large (> 5 km diameter), Quaternary, silicic (>63 wt%

SiO2) calderas in arc settings. Generally, arcs associated with trench-normal convergence rates >70 mm/yr are more likely to host silicic calderas, indicating that a larger flux of basalt favors formation of evolved magma, though notable exceptions such as New Zealand and the Marianas Islands exist. The nature of the crust controls the types of calderas that develop; large and rhyolitic calderas tend to occur on old, thick, continental crust near pre-existing structures and/or areas under local extension.

In continental margins, silicic calderas are distributed more widely behind the volcanic front than are typical arc volcanoes. In addition, arcs with greater caldera densities

7 tend to be young or have migrated, suggesting that the overlying crustal column can be depleted of a low-melting-temperature component over time.

These results allow geologists studying batholiths to use silica content, size, and distribution of plutons to infer the tectonic properties of ancient arcs. This research likewise has implications for volcanic hazard assessment by indicating those regions most likely to have large and/or rhyolitic caldera-forming eruptions—those that are particularly explosive—in the future.

2.2 Introduction

Much attention has recently been given to silicic, caldera-forming eruptions from both scientific and potential hazard perspectives (e.g., Sparks et al., 2005; Geyer and Martì, 2008). A deep scientific understanding of the development of large-scale silicic magmatism is of fundamental importance for studying continental growth and interpreting relict arcs. Continental crust grows primarily by the addition of basalt from the mantle, but evolves and stabilizes by the formation and emplacement of more silicic magmas into the upper crust (Rudnick, 1995). This reorganization of material occurs at both convergent plate margins and in regions of continental extension.

Silicic magmatism in volcanic arcs is of particular interest because the creation of low- density material within continental margins ensures that continents continue to grow rather than be subducted and recycled. Large-scale (>5 km3) silicic magmatism in ancient arcs is represented by granitic and granodioritic plutons, whereas in active arcs silicic calderas are the primary evidence for voluminous silicic magma bodies. Recent work has focused on ignimbrite “flare-ups” (e.g., Ducea and Barton, 2007; de Silva and Gosnold, 2007), which are characterized by the formation of multiple silicic

8 calderas in a concentrated region over a few million years, exemplified by the

Altiplano-Puna complex of the central Andes during the Tertiary. Establishing a baseline for what might be considered “normal” silicic magmatism in volcanic arcs, both in terms of abundance and tectonic setting, would be valuable for studying past ignimbrite flare-ups. Additionally, many relict arcs, such as the Sierra Nevada batholith of North America, are defined by exposed plutons locally intruding silicic metavolcanic rocks of similar age. Establishing relationships between tectonic setting and the compositions, volumes and spatial distributions of silicic calderas would enable geologists to better interpret the original tectonic settings of exposed plutons and batholiths.

Very large caldera-forming eruptions, or “super-eruptions” of more than 1000 km3 (Miller and Wark, 2008), such as the Youngest Toba Tuff of Sumatra, are fortunately rare, estimated to occur only once about every 10,000 years worldwide

(Sparks et al., 2005). But smaller caldera-forming eruptions (CFEs), such as Krakatau,

Indonesia, occur more frequently, on the order of about every hundred to thousand years. While it is thus not surprising that there has not been a super-eruption during historical time, it is possible that one will take place in the foreseeable future, and smaller CFEs will continue to punctuate human history. Despite the low probability of such eruptions, their potentially catastrophic effects necessitate that these types of magmatic systems be studied and understood.

This work seeks to elucidate the occurrence, characteristics, and distribution of silicic calderas in Quaternary volcanic arc settings through a global compilation. We have amalgamated detailed geological and tectonic information for more than 100

9 silicic calderas from all currently active volcanic arcs in a database. This information is used to examine (1) the relationships between the composition and size of arc calderas and their respective tectonic and crustal settings; (2) the differences between the relative abundance of silicic calderas in modern arcs; and (3) the across-arc spatial relationship between silicic calderas and more typical manifestations of arc volcanism.

While recognizing that not all large silicic eruptions are accompanied by caldera collapse, as there are rare silicic arc lavas that exceed 5 km3, this work focuses on silicic caldera formation for two reasons. First, calderas represent large surficial features that, in comparison to isolated domes or lava plateaus, can be easily recognized and are thus very likely to be studied even in regions that are difficult to access, if only at a cursory level. Second, the hazard presented by these explosive eruptions that lead to caldera collapse is considerable, given the great speed of emplacement of pyroclastic flows and the possible global effects if the associated eruptive column is large enough to release a significant volume of ash and volcanic gases into the stratosphere.

2.3 Background: Choosing the Examined Parameters

Based on an initial investigation, we noted that arcs with different tectonic properties had dramatically different quantities of silicic calderas, motivating this study of tectonic effects on caldera abundance. The tectonic parameters we examine here were chosen based on assertions in previous studies that they influence magma chamber evolution and should thus show clear relationships with caldera composition and size.

10

The thickness and nature of underlying crust have been cited as important factors in developing silicic magmas, based both on observation (Hildreth and

Moorbath, 1988; Hughes and Mahood, 2008) and physical modeling (Annen et al.,

2006). Thick, felsic, continental crust is thought to favor formation of evolved magmas because a thicker crustal column consisting of lower-density, low-melting- temperature material will (1) favor the stalling and fractionation of parental magmas during ascent, and (2) increase the likelihood of crustal interaction through partial melting of the crust.

Extensional stresses have also been cited as favoring silicic magmatism, again, both from observation and modeling. Regions of extension, such as northern Central

America (Carr et al., 2003), and thickened compressional regions that have undergone post-orogenic collapse, such as the central Andes (Riller et al., 2001), both host silicic calderas. Physical modeling (e.g., Segall et al., 2001; Jellinek and DePaolo, 2003) demonstrates that a moderate amount of upper crustal extension actually favors magma chamber growth over dike propagation to the surface, because the effective driving pressure of the magma is diminished by extensional stresses. On the other hand, thermal modeling indicates that arcs under high compression associated with crustal thickening will be associated with elevated crustal temperatures over a wide region because of the progressive thickening of heat-producing crust (Pope and

Willett, 1998).

When examining individual calderas or arcs, researchers have frequently noted that calderas are co-located with large structural features such as graben and strike-slip zones (e.g., Erlich, 1986; Bellon et al., 2004), and the locations of some calderas are

11 related to smaller-scale faulting. Lake Atitlan, Guatemala, for example, is intersected by two strike-slip faults (Newhall, 1987). Pre-existing zones of weakness along which magma could be repeatedly supplied to the same location may favor magmatic accumulation and differentiation. Based on geophysical modeling, magma will intrude along pre-existing fractures given a combination of favorably oriented regional stresses and a high magmatic pressure (Delaney et al., 1986).

Numerical modeling of basaltic input in arcs indicates that a minimum magmatic intrusion rate of 50 m (vertical extent) per 10,000 years (a proxy for flux) to the same location is necessary to generate a silicic magma chamber by magma fractionation and partial melting of the crust (Annen and Sparks, 2002). In arc settings, the rate of magma supply within a given arc is a direct function of the associated convergence rate (Hochstein, 1995; Clift and Vannucchi, 2004). Because the rate of plate motion below the mantle wedge is of primary importance, the trench- normal convergence rate is considered here.

Given that arc volcanism is associated with the dynamics of the subducting slab and its interactions with the overlying mantle wedge, we consider slab dip and slab age to be potential factors in controlling the relative abundances of silicic calderas. It has also been noted that oblique subduction may relate to caldera formation in specific settings where transverse motion of the subducting plate is partitioned onto arc-parallel strike-slip faults that host calderas. One such example is

Sumatra where oblique subduction has created the Great Sumatran Fault, a zone containing at least three large silicic calderas, including Toba, which was the site of

Earth’s youngest super-eruption at 76 ky (Bellon, 2004).

12

Several authors have noted that silicic calderas in arc settings frequently occur far from the volcanic front (e.g., Mexico (Ferriz and Mahood, 1984), Central America

(Carr et al., 1990), Southern Andes (Hildreth et al., 1999)). Previous research focusing on arc width without considering different types of volcanism has demonstrated that arcs form ribbon-like two-dimensional shapes, rather than straight lines (de Bremond d’Ars et al., 1995). We here compare the distribution of silicic calderas behind the volcanic front to the distribution of more typical arc volcanoes.

Timescales on the order of 105 – 106 years of consistent magmatic emplacement are necessary to generate the large magma bodies involved in CFEs

(Jellinek and DePaolo, 2003). How this time dependence applies to entire volcanic arcs is, however, unclear. Do all arcs evolve towards more silicic compositions over time, as has been proposed for island arcs (Baker, 1968)? Can arcs that have remained in the same location deplete the lithosphere of easily fusible material and fluids that contribute to partial melting? The duration of each arc was thus recorded in order to determine whether sustained volcanic arc activity is associated with a greater (or smaller) proportion of silicic calderas.

2.4 Data and Methods

2.4.1 Data Specific to Each Caldera

In order to determine the effects of tectonic and crustal setting on the size and composition of individual calderas, data relating to each caldera and its geological location was quantified. Silicic, here, refers to those calderas that have CFE-related ignimbrites with greater than 63 wt% SiO2. Silicic calderas included in this study

13 were limited to those younger than 2 Ma in order to ensure that the present tectonic setting is approximately equivalent to its state during caldera formation. Only calderas larger than 5 km in averaged diameter, formed by a caldera-forming eruption, were considered so as to exclude misnamed sector collapses and explosively excavated craters. The data set for this study builds on that presented in Hughes and Mahood

(2008), and includes several additional arcs and arc sections, as well as a few amendments as new information has become available. The list of included calderas and values of tectonic parameters was developed from extensive regional literature review in addition to previous compilations, including the Newhall and Dzurisin 1988 caldera compilation, the Smithsonian Global Volcanism Program database (SGVP,

Siebert and Simkin, 2002), and the Collapse Calderas Database (CCDB, Geyer and

Martì, 2008).

For each individual caldera, we completed a detailed literature review from published papers, theses, maps, and, when necessary, conference proceedings and personal communication. From this body of information we compiled the age and composition of each CFE, as well as the averaged diameter of the associated ring fracture. Volumes of CFEs were also recorded, but we focus on caldera diameter because volume estimates were not available for about one-third of the calderas, and calculated erupted volumes are often inaccurate in poorly exposed regions.

Compositions were divided based on the highest SiO2 wt% recorded for the CFE: dacitic (63-68 wt% SiO2), rhyodacitic (68-72 wt% SiO2), and rhyolitic (>72 wt%

SiO2). For those calderas with multiple CFEs, the characteristics of the oldest ignimbrite were used when possible. Rhyolitic and large (≥15-km-diameter) calderas

14 are of particular interest as they respectively represent the most evolved end member of volcanism and the extrusive equivalents of batholith-sized bodies.

In addition, a geological and structural description was completed for each caldera that includes the arc “type”, oldest crustal age, crustal thickness, the presence or absence of pre-existing structures, and a characterization of the local state of stress.

In comparing different arc “types”, we have adopted and adjusted the arc categorization developed by Geyer and Martì (2008)1, dividing arcs into oceanic, transitional, or continental based on the underlying crust. For those arcs that are transitional or continental, crustal age is defined as the oldest age of the underlying crust (Tertiary, Mesozoic, or Paleozoic-Precambrian). Because of accretionary processes, very few volcanic arcs actually lie on Precambrian crust, and we have thus grouped those regions with Paleozoic crust.

The presence of pre-existing structural features (including large, arc-wide features and small isolated faults) and the local stress regime (extension, transtension, transpression, or compression) were determined by reviewing region-specific geological descriptions and maps or, when no published data were available, by consulting the CCDB (Geyer and Martì, 2008), the World Stress Map (Heidbach et al.,

2008), or Newhall and Dzurisin’s 1988 compilation. For large regional features, we took a liberal definition of co-location, wherein a caldera is considered to be associated with a regional structure if it lies within or on the margin of the broad area

1 Geyer and Martì (2008) further divide arcs into thick and thin crust. Given an almost total lack of volcanic arcs with “thin” (<10-15 km) oceanic crust, and an abundance of intraoceanic arcs categorized as transitional “thin” (<20-25 km), we grouped Geyer and Martì’s oceanic “thick” and “transitional thin” together as oceanic.

15 defined by an established graben, fault zone, or fold-and-thrust belt. It should be noted that these arc-wide features are typically much larger than the 5- to 20-km diameters typical of the studied calderas. For smaller isolated structural features, such as individual strike-slip faults, the feature must intersect the caldera in order to be considered co-located.

2.4.2 Data Specific to Arcs and Arc Sections

Arcs were defined as regions with volcanoes related to subduction, and were divided into arc sections based on differences in underlying crust, sharp changes in subduction tectonics, and previous studies. In order to statistically compare the impact of different tectonic parameters on caldera formation, arcs and arc sections were divided into 100-km segments, and each segment was assigned values of subduction- and arc-related parameters, including convergence rate, the dip and age of the descending slab, obliquity of subduction, upper plate stress regime, and arc duration.

In addition, a crustal thickness and an arc “type,” as defined above, were recorded for each segment.

Arc durations were gleaned from regional studies of individual arcs or arc sections. Arc duration was examined in two ways: as the longevity of subduction- related volcanism in a given arc (total arc duration), and the time span of arc volcanism in its current location (current arc duration). For arcs that have not migrated these two measures are equal.

Parameters related to subduction, including trench-normal convergence rate, slab dip, subduction obliquity, and upper plate stress state, were taken from subduction

16 zone transects included in recent compilation studies (Lallemand et al., 2005, Cruciani et al., 2005). Data for arcs missing from these compilations were gathered from literature review and the HS3-NUVEL 1A plate model (Gripp and Gordon, 2002).

Given that higher crustal temperatures should favor silicic magma formation, and the modeled connection between elevated crustal temperature and compression in arcs (Pope and Willett, 1998), we have chosen to look at the overall stress regime of the upper plate, as defined by Lallemand et al. (2005). End-member definitions

(backarc thrusting and backarc spreading) are based on geological observation, whereas intermediate values (ranging from moderate extension to strike-slip to moderate compression) were based on focal-mechanism analysis. It is important to note that the stress regime of the upper plate and the local stresses acting on the caldera-forming magma chamber may differ. One example of this phenomenon is in the Ecuadorian Andes, a region of 50-km-thick continental crust that is characterized as compressional, but locally experiences tension, manifested as a graben between two

Cordillera.

2.4.3 The Control Group: “Normal” Arc Volcanoes

In order to compare the abundance of silicic calderas with a control group, we also compiled a list of major arc volcanoes, primarily from the SGVP, which are defined here as stratovolcanoes, shield volcanoes, complex volcanoes, and non-silicic or small (< 5-km diameter) calderas. Smaller monogenetic volcanic features such as cinder cones, lava domes, and maars are not considered given their very short lifetime in the geological record and association with small volumes of magma.

17

2.4.4 Across-arc Distribution of Silicic Calderas

In order to compare the spatial distribution of silicic calderas behind the volcanic front to the distribution of more typical arc volcanoes, we determined the distance of each volcanic feature from the volcanic front. An automated program was used to calculate the perpendicular distance between each volcanic feature and the associated trench segment parallel to each 100-km arc segment. Within each 100-km arc segment, the volcano or caldera closest to the trench was declared the “volcanic front,” and its distance from the trench was subtracted from all other feature distances to determine their respective distances from the front (Figure 1). Arc volcanoes included in this part of the study are defined as above. Smaller, ephemeral features are again not considered, as it is likely that these features would tend to be quickly overprinted within the geological record, making them of little use for defining a volcanic front. Slab dip may also be an important factor to consider given that volcanism is generally associated with the surficial region associated with slab depth of approximately 100 km. Calderas farther from the volcanic front might thus be expected to be associated with shallow slab dips.

2.5 Results and Discussion

Characteristics of a total of 108 Quaternary silicic calderas from 29 volcanic arcs and arc sections were compiled in this study (Figure 2). We present our results in three parts. In the first, we examine two caldera attributes, averaged diameter and

CFE composition, with respect to crustal-tectonic setting. Second, tectonic factors are evaluated as to whether they favor silicic caldera formation. Finally, distances from

18 the volcanic front are analyzed, comparing silicic caldera distances to those of more typical arc volcanoes.

2.5.1 Caldera Attributes

The examined caldera dataset contains 34 dacitic, 33 rhyodacitic, and

41 rhyolitic calderas. Average diameters range from 5 km to 45 km with a median diameter of 11 km. The majority of calderas are relatively small, with quantities decreasing with increasing diameter (Figure 3A). Division of the calderas into three equally sized groups based on diameter reveals that small calderas (5-8.5 km) are dominantly dacitic, whereas large calderas (>15 km) are dominantly rhyolitic (Figure

3B).

2.5.2 Crustal Attributes

2.5.2.1 Age, Composition, and Thickness of Crust

Dividing calderas based on the nature of the underlying arc crust (Figure 4 and

Figure 5), we find a dependence of CFE composition and caldera size on the nature of the overriding plate. Although there are calderas of all three CFE compositions in all types of margins, large and/or rhyolitic calderas are more common on old (≥Paleozoic) continental crust, whereas young (Tertiary) or oceanic crust is dominantly host to small dacitic calderas. It should be noted that the total number of silicic calderas in each setting is approximately proportional to the total number of volcanoes in each setting (Table 1). In all, these findings support the assertion by previous studies that old, silicic crust favors formation of highly silicic magmas.

19

The crustal thickness underlying examined calderas ranges from 15 km in the

Kermadec arc to 70 km in the central Andes. If crustal thickness is divided into large bins (thin ≤30 km, normal = 30-40 km, and thick ≥40 km), dependence of CFE composition on crustal thickness is apparent. Settings with thick crust host mostly rhyolitic calderas, whereas thin crust hosts dacitic calderas. The exceptions to this rule demonstrate that crustal thickness alone does not determine CFE composition. For example, New Zealand has several very large rhyolitic calderas but is located on continental crust that is only about 20-km thick (Stratford and Stern, 2006). It has been hypothesized that the highly evolved magma erupted in this region develops by melting of a greywacke protolith in addition to fractional crystallization (Charlier et al., 2005). At the other extreme, although dacitic calderas are largely absent on thick crust, the Altiplano-Puna complex in the Central Andes, located on crust that is 65-70- km thick, contains calderas only of dacitic to rhyodacitic composition. The products from the largest Quaternary caldera in this arc, the Purico complex, are similar to the crystal-rich dacites erupted in this region during the Tertiary “ignimbrite flare-up”

(Schmitt et al., 2001), which are thought to represent magma chambers that erupted before significant segregation of crystal-poor rhyolitic melt (Bachmann and Bergantz,

2008).

It is clear from these results that crustal interaction plays an important role in magmatic evolution and caldera formation, but it is difficult to determine whether the observed silicic compositions are due to assimilation of evolved crust or to mafic magmas stalling and fractionating during ascent through low-density material. There is extensive geochemical and isotopic evidence in several volcanic systems that both

20 processes play a role, and it is likely that the interplay between the two depends on tectonic location. It has been suggested, for example, that small rhyolitic calderas in the Izu-Bonin arc are related to silicic magmas that form by partial melting of calc- alkaline dioritic intrusions in the upper to middle crust (Tamura and Tatsumi, 2002).

In contrast, high-silica rhyolite products from the 11-km-wide La Primavera caldera in

Mexico are thought to have formed by extensive fractional crystallization of a mantle- derived magma that may also have partially melted mafic intrusive rocks in the lower crust (Mahood and Halliday, 1988). We conclude that the nature of the crust underlying a given caldera has a substantial impact on the composition and volume of silicic magma produced, but how different crustal characteristics such as thickness and composition interact to favor crustal assimilation or crystal fractionation may vary considerably, and is likely influenced by the local stress regime in the crust.

2.5.2.2 Pre-existing Structures and Local Stresses

The local stress regime also shows a relationship with both caldera composition and size (Figure 6). Approximately 70% of examined calderas lie in locally extensional environments, and the majority of rhyolitic calderas (85%) and large calderas (89%) reside in extensional settings. Almost two-thirds of the examined calderas are associated with regional structures such as fault zones, graben, and sutures, or intersect small features such as local strike-slip faults and fractures (Table

2). About one-quarter of all examined calderas lie at the intersection of multiple structures. More than 90% of rhyolitic calderas are associated with structural features, and fifteen of the seventeen silicic calderas more than 20 km in diameter were

21 associated with large, arc-wide features. Most of these structures are reflections of large-scale extensional tectonics, given that the dominant type of associated regional feature is a graben (Table 2). The possibility that pre-existing structures and local extension simply favor caldera-forming eruptions over eruption-free magma solidification is impossible to rule out. The fact that caldera diameter appears more dependent on stress regime than does composition (Figure 6) may indicate the role of extension in large-scale caldera formation. The link, however, between large and rhyolitic calderas and local stresses and structures suggests that they contribute to the development of large magma chambers with high-silica compositions.

2.5.3 Tectonic Controls on Silicic Caldera Formation

In order to determine how tectonic setting controls the development of silicic calderas, we devised a measure of the relative abundance of silicic calderas in different arcs. Caldera abundance was examined in a variety of ways to determine which method was most appropriate. We settled on caldera density, defined as the number of examined calderas per 1000 km of arc length. Methods other than caldera density included cumulative caldera area (the total ellipsoidal areas of all arc calderas per 1000 km of arc length) and caldera-volcano ratio (the ratio of silicic calderas to non-caldera arc volcanoes). The results for each abundance type were approximately the same, although the caldera density metric showed a tighter distribution, and allowed for the inclusion of New Zealand on graphs without resorting to a log scale.

Caldera density was compared to tectonic parameters averaged over each arc or arc section (except subduction obliquity). In order to determine whether more

22 typical arc volcanoes exhibited similar relationships to tectonic parameters, volcano density was also calculated, defined as the number of large arc volcanoes (not including silicic calderas) per 1000 km of arc. Additionally, given that subduction parameters can vary over the extent of an arc, we chose to use a nonparametric signed- rank test to statistically compare tectonic parameters for 100-km arc segments that contain silicic calderas with segments that do not.

Oceanic arcs tend to have low caldera densities, whereas those with the highest caldera densities are transitional to continental (Figure 7). There is a positive albeit scattered correlation between caldera density and volcano density (Figure 7A) indicating that, as one would expect, arcs containing more volcanoes likewise tend to host more silicic calderas. But the variation in caldera density observed for any given volcano density reveals that the occurrence of silicic calderas in arcs is not simply a constant function of the number of more typical arc volcanoes. It should be noted that the Taupo Volcanic Zone of New Zealand (arc number 16) lies far above the rest of the data because of its relative lack of intermediate to mafic volcanism in comparison to its voluminous silicic caldera activity.

2.5.3.1 Trench-Normal Convergence Rate

The plot of caldera density versus convergence rate is essentially flat up to ~70 mm/yr, at which point it steepens sharply (Figure 7B). New Zealand (arc number 16) is again clearly an outlier. By comparison, the plot of volcano density versus convergence rate (Figure 7C) is flatter and highly scattered, although volcano density likewise generally increases with increasing trench-normal convergence rate. For both

23 calderas and other arc volcanoes, those arcs with backarc spreading (i.e., those in which new oceanic crust is being formed behind the arc) fall far below the general distribution, regardless of convergence rate. We thus conclude that the relative absence of silicic calderas in these arcs is likely related to an overall scarcity of volcanism in regions with backarc spreading. Setting aside New Zealand and those arcs with backarc spreading, it is difficult to determine whether the graph of caldera density shows a smooth exponential increase with increasing convergence rate, or whether there are two different arc populations: one group at convergence rates less than 70 mm/yr with low caldera densities, and a group with high caldera densities at convergence rates greater than 70 mm/yr. In either case, there is an obvious interplay between caldera density, convergence rate, and the presence or absence of backarc spreading.

A statistical comparison of trench-normal convergence rates (Figure 8) for those segments containing silicic calderas with those that do not shows that the distributions are different at the 94% level of significance using a nonparametric signed-rank test. (Given the skewed nature of the distributions, we chose to use a nonparametric test to compare all distributions rather than assuming normality, as would be necessary for a t-test.) Both plots show rates ranging from about 10 to 150 mm/yr, with maxima in the 70 - 80 mm/yr bin, and very few segments associated with convergence rates higher than 80 mm/yr. The histogram of caldera segments, however, has a far more pronounced maximum, with approximately 35% of the segments falling within the 70 - 80 mm/yr bin, more than twice the quantity in any other bin.

24

Our results indicate that magma supply is a critical factor in controlling whether silicic calderas form in arcs. Assuming that the volumetric rate of magma intrusion below the arc is directly related to trench-normal convergence, the positive correlation between convergence rate and caldera density demonstrates that silicic calderas are far more likely to develop in arcs with a higher flux of mantle-derived magma. The comparison with the plot of volcano density versus convergence rate demonstrates that high magmatic flux is likely to be more important for silicic caldera formation than it is for the development of more typical arc volcanism. Flux may be particularly important for large silicic systems for two non-exclusive reasons: (1) If fractionation were the sole mechanism for generating evolved magmas, producing rhyolite by about 90% fractionation of basalt would require huge volumes of basalt to generate the volumes of rhyolite observed in caldera-forming eruptions; (2) High magma flux adds more heat to the crust, resulting in partial melting of the crust and/or previous intrusions. Given that we do not see large volumes of silicic magma in

Hawaii, which has an extremely high basaltic flux (1.5 x 10-2 km3/yr), magmatic interaction with felsic crust, as a density filter and/or as a source of evolved material, must be important in generating large silicic magma chambers. The absence of a relationship between trench-normal convergence rate and the composition of caldera- forming eruptions (Figure 9) suggests that although a high mantle-derived flux is necessary to generate a large body of silicic magma, it does not favor formation of higher-silica magma bodies (i.e., rhyolitic as opposed to dacitic). We thus favor crustal interaction as the primary mechanism for magmatic evolution to rhyolite.

25

That those arcs with backarc spreading have very low caldera densities may likewise be related to magma supply. Thermal modeling suggests that the presence of a spreading backarc induces a change in corner flow and results in suppression of magmatism near the volcanic front (Conder et al., 2002). The fact that arcs with backarc spreading also have very low volcano densities supports this hypothesis. New

Britain represents an interesting data point in this respect because, unlike other arcs with backarc spreading, it has transitional rather than oceanic crust, and the associated

Manus spreading ridge is relatively short compared to the length of the volcanic arc.

These circumstances may explain why the New Britain arc (arc number 12) plots so much closer to the general curve than other arcs with backarc spreading (Figure 7).

One inevitable question is whether there are submarine silicic calderas that have yet to be discovered in such settings. Although this is possible, it seems unlikely that, for example, the Marianas (arc number 11) have enough undiscovered silicic calderas to raise the caldera density to a value predicted by the trend line in Figure 7 for that convergence rate.

2.5.3.2 Subduction Obliquity

Examination of individual arc segments allows analysis of subduction obliquity, the measure of the angle between a line normal to the trench and the direction of convergence for each segment. This parameter can vary dramatically over any given arc and is thus not averaged. Comparing arc segments containing silicic calderas with those that do not, both silicic calderas and volcanism in general tend to decrease with increasing subduction obliquity (Figure 10). This is likely related to the

26 diminishing of trench-normal convergence rate (and associated magma flux) with increasing obliquity.

2.5.3.3 Arc Duration

There is no observable relationship between total arc duration and either caldera density or volcano density (Figure 11A, C). More striking are the wedge- shaped plots of caldera and volcano densities versus current arc duration (Figure 11B,

D), i.e., the duration of the arc in its current geographic location. For both calderas and other arc volcanoes there is a large spread in density values at short durations, whereas at longer durations, densities are limited to small values. Arcs that have a caldera or volcano density above the fiftieth percentile have all migrated or have been active for time periods of less than thirty million years. A possible explanation for this observation is that arcs that remain in a single location use up easily fusible material and/or volatile enrichment from one portion of the mantle wedge or overlying lithosphere, resulting in a decrease in the rate of magma supply over time. The fact that this wedge-shaped distribution is observed for both caldera density and volcano density (which includes large basaltic shield volcanoes) argues against an explanation based on crustal attributes that would limit magma storage or evolution. Given the large range in the data, however, arc duration is a factor secondary to convergence rate in determining caldera abundance.

Looking at average CFE composition as a function of arc duration (Figure 12), there is not a clear relationship between either measure of arc duration and average

CFE composition. The very oldest arcs (total duration > 150 Ma) tend to have

27 dominantly rhyolitic CFEs and the strictly dacitic arcs tend to be younger than 60 Ma.

These observations are likely related to arc type, in that continental arcs, which tend to host rhyolitic calderas, also tend to have the longest durations.

2.5.3.4 Slab Age and Slab Dip

In order to assess whether the age and dip of the down-going slab influence the abundance of silicic calderas, we plotted the arc-averaged values of these parameters against caldera density (Figure 13A, B) and we compared these parameters for arc segments with and without silicic calderas (Figure 14A, B). When averaged over each individual arc, slab age appears to be unrelated to caldera density (Figure 13A). The slab-dip plot, on the other hand, shows a peak in caldera density at intermediate values

(Figure 13B). This may indicate that there is a slab dip between 30o and 40o associated with a tectonic setting that favors silicic caldera formation, although even at these intermediate values there are arcs with very low caldera densities. Comparing slab dip for arc segments with and without calderas (Figure 14B), there is a statistically significant difference (using a nonparametric signed-rank test) between the distribution of slab dips for all segments and those segments hosting silicic calderas.

The distribution of the segments with calderas is highly skewed, with more than 40% of segments containing calderas having a dip angle between 30o and 40o. The abundance of silicic calderas with slab dip angles of 30-40o may be in part related to the observation that continental arcs are generally associated with a shallower range of slab dips than are oceanic arcs, and transitional arcs span a range of dips between the two (Lallemand et al., 2005; Cruciani et al., 2005). Because several oceanic arcs have

28 few silicic calderas (Figure 7), these relationships are reflected in the histograms of slab dip.

2.5.3.5 Thickness of Crust

Similar to slab dip, the plot of caldera density versus crustal thickness (Figure

13C) shows a peak at intermediate values of 30-40 km. This is the most common crustal thickness for all arc segments (Figure 14C), but especially so for those hosting calderas, as about 45% of arc segments with silicic calderas have crust 30-40-km thick. However, a statistical comparison of crustal thickness for arc segments containing calderas with those that do not shows that there is not a statistically significant difference between the two populations.

2.5.3.6 Upper-plate Stress Regime

The histogram of upper-plate stress regimes of arc segments (Figure 14D) reiterates the lack of silicic calderas in arc settings with active backarc spreading. In addition, this plot shows that arc segments in which the upper plate is under compression are slightly more likely to contain calderas than are those in which the upper plate is undergoing tension or shear.

2.5.4 Distance from Volcanic Front

When all silicic calderas are examined as a group, the distribution of their distances from the volcanic front is statistically indistinguishable from the distribution of other major arc volcanoes (Figure 15A). But when silicic calderas are divided into groups based on arc type, we observe that silicic calderas in continental margins are

29 not concentrated along the volcanic front, but are instead distributed more widely behind the front (Figure 15D). Using a signed-rank nonparametric test, we calculate that the caldera and volcano distributions in continental margins are statistically different at the 95% significance level (p = 97.7%). It is apparent that silicic calderas in continental arcs are not as tightly grouped at the volcanic front, as are more typical arc volcanoes.

This comparatively wide distribution of silicic calderas behind the volcanic front in continental margins may have a variety of explanations. First and foremost, arcs located in continental margins have a much broader area of evolved crust in which “MASH” processes of crustal assimilation and partial melting of previous intrusions (Hildreth and Moorebath, 1988) are likely to occur. By comparison, oceanic arcs have a relatively narrow band of more evolved crust with which magmas can interact, as seen in the Izu-Bonin arc where a number of small silicic calderas lie directly on the volcanic front. While this explanation accounts for why silicic calderas are more widely distributed in continental margin arcs than in oceanic arcs, it does not explain why we observe no difference between volcano and caldera distributions on transitional crust, which is frequently subaerial over a broad region, similar to continental arcs (e.g., northern Japan).

There appears to be no correlation between slab dip and the distance of individual calderas behind the volcanic front (Figure 16). Shallower dips do not necessarily correlate with calderas farther from the volcanic front and, likewise, arcs associated with steeply dipping slabs commonly have silicic calderas far from the

30 volcanic front. This agrees with previously published results showing that arc width does not correlate with slab dip (de Bremond D’Ars et al., 1995).

2.5.5 Association With Structural Features

We propose that the observed differences in typical arc volcano and silicic caldera distributions are related, in part, to variations in the quantity and location of pre-existing structures in different types of arc settings. In order to test this hypothesis, we examined whether silicic calderas in specific arc settings were associated with structural features (Figure 17A), and whether silicic calderas associated with structural features had a significantly wider distribution behind the volcanic front (Figure 17B). Silicic calderas located on overriding crust that is continental are most likely to be associated with structural features, and calderas on transitional crust are more likely to be associated with pre-existing structures than those on oceanic crust. The greater influence of structural control on the location of calderas in continental arcs seems likely to be a result of protracted histories that includes terrane accretion, rifting, and strike-slip fault zones. Additional evidence for the role of structural features comes from the observation that calderas associated with pre-existing structures tend to be spread over a wider area behind the volcanic front

(Figure 17B), although a statistical test shows that the likelihood of the distributions being from different populations is only 75%.

31

2.6 Conclusions and Implications

2.6.1 Conclusions

Based on this global compilation we conclude the following about silicic calderas in arc settings:

1) The nature of the underlying crust has a profound impact on caldera size and

the composition of the caldera-forming ignimbrite. Although calderas of all

compositions and sizes occur in all observed continental settings, large (≥15

km), rhyolitic calderas are dominantly located on evolved, continental crust

that is Paleozoic or older.

2) Local extension and the presence of structures such as fault zones, graben, and,

to a lesser extent, small-scale faulting tend to favor higher-silica compositions

of the caldera-forming unit. The majority of individual calderas (65%) were

located in areas of local extension and proximal to pre-existing structural

features. In particular, the vast majority of large and rhyolitic calderas were

located in such settings. However, in comparing the regional upper-plate stress

regime of arc segments that host silicic calderas to that of all arc segments,

segments hosting calderas are slightly more likely to be under compression and

much less likely to be undergoing backarc spreading.

3) Magmatic flux is the primary factor in determining the abundance of silicic

calderas in a given arc. In arc settings, this flux is related both to the trench-

normal convergence rate and to the presence or absence of backarc spreading,

which may lower the amount of magma supplied to the volcanic front.

32

4) The duration of arc activity in a specific location is a secondary factor in

controlling caldera and volcano abundance. Arcs that have remained in the

same position for more than twenty million years tend to have lower caldera

and volcano densities, although the converse is not necessarily true. Arcs with

high caldera densities have either migrated during the last thirty million years

or been active for less than this amount of time.

5) In continental margins, silicic calderas are spread over a wider region behind

the volcanic front than are more typical arc volcanoes. This is likely related to

the fact that continental margins, unlike oceanic margins, have a wide region

of low density material that can be assimilated and stall magmatic rise,

resulting in more evolved compositions. In addition, because of their

protracted histories, continental margins host many pre-existing structures that

favor magma chamber growth and evolution.

2.6.2 Further Implications

2.6.2.1 Extensional Settings

Our conclusion that extension favors silicic caldera formation would seem to contradict our observation that arcs with backarc spreading actually have very little arc volcanism. This apparent contradiction might be explained if we assume that once intra-arc extension overcomes a threshold, the underlying crust would be thinned to the extent that new oceanic crust could form (e.g., the Marianas). New Zealand may be a location where local extension and crustal thinning will eventually produce a fully rifted arc.

33

2.6.2.2 Ignimbrite Flare-ups

Among all the modern arcs we studied, New Zealand is unique. The volumetric eruptive rate in the Taupo Volcanic Zone is much higher than expected given its local rate of subduction. The anomalously high caldera density, which at twenty-three is nearly ten times the median for all other arcs, qualifies Taupo as a modern “ignimbrite flare-up”. For comparison, we calculated caldera densities for the

Tertiary “ignimbrite flare-up” in the Altiplano-Puna of the Central Andes (based on maps from deSilva and Gosnold, 2007), which were on the order of 12-14 for each 2-

Ma period. This should be considered a minimum, given that the volumes erupted were very large, about 1.7 x 103 km3/Ma, and smaller features may be missing from the geological record. Given that this extrusion rate is slightly lower than the 2.4 x

103 km3/Ma observed at Taupo (Houghton et al., 1995), we conclude that it can be considered an “ignimbrite flare-up,” although the mechanism for this event is unclear.

The mass flux over the past 2 Ma in the Taupo Volcanic Zone is comparable to that observed at Yellowstone, but Taupo has far more frequent small eruptions potentially because it lies on highly extended, thinned continental crust (Houghton et al., 1995). Unlike Yellowstone and the Altiplano-Puna, respectively, New Zealand cannot be related to a mantle plume (Hochstein, 1995) or delamination of an overly thickened lithosphere. It is possible that the observed silicic activity is associated with an early stage of arc rifting that involves rapid extension and thinning of the lithosphere resulting in mantle upwelling that promotes partial melting of a fertile greywacke crust.

34

2.6.2.3 Interpreting Relict Arcs

The observations and trends presented here should be advantageous in interpreting the tectonic settings of relict arcs. Eroded arcs that consist of plutons the upper reaches of which are dominantly granodioritic, small (less than 8.5 km in diameter), and having a relatively linear along-strike distribution, are likely to have formed in island arc settings on mafic crust. Ancient arcs that contain large granitic plutons that are distributed over a broad area are likely to have formed on continental crust. The polarity of ancient subduction can be determined from the observation that silicic magmatism is distributed further from the volcanic front than is the more typical mafic and intermediate volcanism of arc. In addition, if plutons have been carefully dated, their densities could be compared to present caldera densities in order to determine whether an anomalously high or low amount of silicic magmatism was present. This “pluton density”, however, would be a maximum given that not all silicic magma chambers need be associated with caldera forming eruptions.

2.6.2.4 Volcanic Hazards

Although we stress that the conclusions from this paper are not intended to pinpoint the most (or least) probable locations of future caldera-forming eruptions, this study does have general hazard implications. Assuming that the recent past is key to understanding volcanic hazards in the future, this study indicates that certain types of arc settings are more likely to host silicic CFEs. Specifically, caldera density tends to be high in arcs with trench-normal convergence rates greater than 70 mm/yr and crustal thickness over 15 km. Moreover, the largest and most rhyolitic eruptions (i.e.,

35 those most likely to be highly explosive) tend to be located on continental crust that is at least Paleozoic in age and thus likely to be more felsic. Conversely, silicic CFEs have been largely absent from oceanic arcs located on thin crust (<15 km) or with active backarc spreading. It is important to note, however, that silicic CFEs have occurred in a wide variety of arc locations, from ocean islands to continental plateaus, and that even a small (5 km3) dacitic CFE could present a significant hazard, depending on its location (proximity to sizable populations, infrastructure, air traffic, etc). In addition, New Zealand represents an anomalous zone that, despite having a relatively low associated convergence rate, has an exceptionally high caldera density.

Thus, no volcanic arc can be categorically ruled out as being able to host a silicic CFE, although certain arcs are currently far more likely candidates than others.

36

2.7 Tables

TABLE 1: TECTONIC SETTINGS OF EXAMINED CALDERAS AND VOLCANOES Other arc Total number of Arc type Silicic calderas * volcanoes examined features

Oceanic crust 22 (21%) 196 (27%) 218 (26%)

Transitional crust 35 (32%) 166 (23%) 201 (24%)

Continental crust 51 (47%) 368 (50%) 419 (50%)

Total 108 (100%) 730 (100%) 838 (100%) Note: Caldera details in Appendix A. Reference details are in Appendix C. * = Does not include silicic calderas. Table 1: Tectonic settings of examined calderas and volcanoes

37

TABLE 2: STRUCTURAL SETTINGS OF EXAMINED CALDERAS

Type of feature Calderas (n=108)

Regional structures

Graben or rift zone 45 (42%)

Fault zone 18 (17%)

Fold and thrust belt 10 (9%)

Any regional structure 67 (62%)

31 (29%) Local small-scale faulting

All structures Any associated structure 69 (64%)

Intersection of structures 26 (24%)

Local stress regime Compression 26 (24%)

Transpression 3 (3%)

Transtension 32 (30%)

Tension 47 (44%) Note: Calderas can have more than one structural feature. Table 2: Structural settings of examined calderas

38

1 3 2 0 3 2 1 - - 1.6 3.0 0.3 0.4 3.0 3.0 1.0 0.0 0.0 2.0 1.0 2.0 3.0 3.0 3.0 2.0 0.0 0.5 1.0 0.0 1.0 3.0

------

stress code Upper plate

30 30 20 34 45 30 35 22 32 23 20 32 28 12 15 20 27 33 40 34 37 35 22 50 70 40 31 30 14 (km)

rustal thickness C

normal convergence rate. Averaged CFE normal convergence rate. Averaged CFE - - 57 81 26 49 50 31 48 58 52 10 13 23 27 19 12 54 38 18 98 37 103 125 129 131 148 107 100 107

Slab age (Ma) Slab

35 40 43 36 60 44 38 39 32 46 60 58 56 46 51 47 62 36 54 31 44 48 37 46 28 37 10 53 62 km

Slab dip at 100 Slab

N convergence rate isN trench -

7 6 4 30 63 42 10 66 43 43 25 23 23 12 55 42 35 38 60 60 50 23 30

145 100 180 185 165 120 (Ma)

Total arc duration Total arc duration

5 7 7 6 5 4 3 2 7 3 20 10 25 13 43 43 25 23 55 42 38 12 12 25 25 26 15 23 120

Current arc Current Table duration (Ma) duration 3:

ude connotesude intensity), 0 indicates or transverse, neutral 1 to 2 is tension, 3 is backarc spreading. Other definitions

- - - - 2.5 3.0 1.2 1.5 3.0 2.2 2.2 1.8 2.2 1.8 1.8 1.0 1.5 3.0 1.0 2.0 2.0 3.0 1.9 2.0 1.0 3.0 1.3 2.3 1.0 composition Average CFE Average CFE

33 38 62 73 73 76 76 86 46 96 55 38 56 54 31 52 71 81 42 66 72 20 10 64 1 is compression (magnit 150 107 167 39.5 23.5 – (mm/yr) 3 to - N convergence rate N - ldera and volcano density of features are number per 1000 km of arc length. T T TABLE 3: AVERAGED TECTONIC PARAMETERS OF EXAMINED ARCS OF EXAMINED PARAMETERS TECTONIC TABLE 3: AVERAGED

8 12 16 30 23 18 21 70 29 39 17 21 23 15 20 10 20 26 32 20 33 63 48 43 36 41 36 21 18 density Volcano continental. Ca -

2 18 12 23 1 4.0 1.8 3.3 2.9 1.3 8.6 9.2 3.3 0.0 6.7 0.9 0.0 1.4 0.8 2.6 2.6 2.1 8.0 2.9 3.3 1.9 4.6 0.0 2.9 0.0 density Caldera Caldera transitional, 3 -

rhyolite. Upper plate stress code: -

3 3 2 2 2 2 3 1 2 1 1 2 2 1 1 3 1 3 3 3 3 2 1 3 3 3 3 1 1 code Arc type oceanic, 2

-

1 2 3 4 5 6 7 8 9 rhyodacite, 3 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 - Arc #

dacite, 2 -

Arc type codes: 1

– land

Bali - Bonin - Arc Name Aegean Sumatra Java PhilippinesN. Philippines S. Ryukyu N. Kamchatka Kurile Japan NE Izu Marianas New Britain Vanuatu Tonga Kermadec New Zea Aleutians Alaska Pen Cascades Mexico America Cen. N. America Cen. S. Panama Andes N. Andes C. Andes S. Austral Andes L. Antilles Sandwich S. Notes: 1 composition: 1 and sources in text. Averaged tectonic parameters of examined arcs

39

2.8 Figures

Figure 1: Method for measuring distance of features from the volcanic front. In this example from a 100-km segment (bounded by dotted lines) in the northern Andes, Quilotoa (Ql) is the closest feature (perpendicular distance) to the trench and is therefore defined as the volcanic front. The behind-the- front distances of calderas (circles) and other arc volcanoes (triangles) are calculated by subtracting Quilotoa’s distance from the trench from all other distances from the trench. Examples are shown for La Chacana (Chc), Chalupas (Chp), and Sumaco (Sm).

Figure 2: Locations of examined arcs and calderas. Arcs are coded by “arc type” based on the nature of the underlying crust. Silicic calderas are represented by circles and are coded by size and the maximum silica content of the caldera-forming eruption. Arcs and arc segments are: 1-Aegean, 2-Sumatra, 3- Java-Bali, 4-N. Philippines, 5-S. Philippines, 6-N. Ryuku, 7-Kamchatka, 8-Kurile, 9-N. Japan, 10-Izu- Bonin, 11-Marianas, 12-New Britain, 13-Vanuatu, 14-Tonga, 15-Kermadec, 16-New Zealand, 17- Aleutians, 18-Alaska Peninsula, 19-Cascades, 20-Mexico, 21-N. Central America, 22-S. Central America, 23-Panama, 24-N. Andes, 25-Central Andes, 26-S. Andes, 27-Austral Andes, 28-Lesser Antilles, 29-S. Sandwich.

41

Figure 3: Summary of silicic caldera diameters and compositions. A) Distribution of caldera diameters. B) Examined calderas divided into three equal groups based on diameter and coded by composition of the caldera-forming eruption (CFE). Small calderas are dominantly dacitic, whereas large calderas are primarily rhyolitic.

42

Figure 4: Relationship between CFE composition and the nature of the underlying crust. See text for definitions of crustal parameters. Calderas of all compositions are found in all tectonic settings. Rhyolitic CFEs, however, tend to be dominant in more evolved crustal settings (i.e., thick, old, continental crust). Dacitic CFEs tend to be more prevalent in regions with less-evolved crust (i.e., oceanic, thin crust). Note, however, that several rhyolitic calderas occur in New Zealand, which has a crustal thickness of only 20-25 km.

43

Figure 5: Relationship between caldera diameter and the nature of the underlying crust. Calderas of all sizes occur in all tectonic settings. The largest calderas (those > 15 km in diameter) are prevalent on more evolved crust (i.e., continental, Paleozoic or older, and thick). The smallest calderas tend to occur in oceanic settings.

44

Figure 6: Relationship between structural setting and (A) CFE composition and (B) caldera diameter. Extensional settings include both tensional and transtensional settings. Compressional settings likewise include compressional and transpressional settings. More calderas are found in extensional structural settings. In particular, a greater proportion of rhyolitic and large calderas are found in extensional settings than in compressional ones.

45

Figure 7: Correlation between caldera density and other arc volcano density (A), relationship between trench-normal convergence rate and caldera density (B), compared to that for all other arc volcanoes (C). Density is defined as the number of features per 1000 km of arc length. Arc segments in regions of backarc spreading, all of which have low caldera and volcano densities, are darkened.

46

Figure 8: Histograms of trench-normal convergence rates for (A) 100-km arc segments containing at least one silicic caldera, and (B) arc segments not containing calderas. There is a pronounced peak around 70-80 mm/yr in the distribution of segments containing calderas that is not present in the distribution of all arc segments. ‘p’ represents the probability that the two distributions are statistically different based on a signed-rank nonparametric test. There is clear evidence that the two distributions are statistically different at the 95% confidence level.

47

Figure 9: Averaged caldera composition versus trench-normal convergence rate for each arc shows no obvious relationship. 1 – Dacite; 2 - Rhyodacite; 3 – Rhyolite. Note that arcs without calderas are not included.

48

Figure 10: Histograms of subduction obliquity for (A) 100-km arc segments containing at least one silicic caldera, and (B) arc segments not containing calderas. The metric ‘p’ represents the probability that the two distributions are statistically different based on a signed-rank nonparametric test. Given this 74% probability, there is not statistically clear evidence that subduction obliquity favors caldera formation.

49

Figure 11: Relationship between the longevity of volcanic arcs and their respective caldera densities (A & B) and volcano densities (C & D). Total arc duration (A & C) is the total time during which volcanic activity associated with subduction has persisted, whereas current arc duration (B & D) is the lifespan of arc volcanism in its present location. There is a slight negative correlation between current arc duration and both caldera and volcano densities. Arcs that have remained in one place for than 30 Ma tend to have fewer silicic calderas than arcs that are relatively young or that have migrated. Tight spacing of data points prevents labeling of arcs but data are listed in Table 3.

50

Figure 12: Relationship between the longevity of volcanic arcs and their averaged caldera compositions. . 1 – Dacite; 2 - Rhyodacite; 3 – Rhyolite. Although there is little relationship between total arc duration and the average caldera composition of a given arc, it is notable that all arcs containing solely dacitic CFEs are younger than 60 Ma. Current arc duration shows no relationship to the composition of a given arc’s calderas. Note that arcs without calderas are not included.

51

Figure 13: Caldera density as a function of arc type and subduction parameters. (A) Age of the down- going slab at the trench. (B) Dip of the down-going slab at 100-km depth, (i.e. beneath the volcanic arc). (C) Average thickness of the crust underlying the arc or arc section. The plots for both crustal thickness and slab dip show a peak of maximum caldera densities at intermediate values. Tight spacing of data points prevents labeling but data are listed in Table 3.

52

Figure 14: Comparison of subduction parameters for 100-km volcanic arc segments containing calderas with all arc segments. (A-C) Parameters as defined in Figure 13. Five arc segments (which do not contain calderas) have unknown slab ages and are thus not included in A. (D) Compression bin includes compressional and transpressional stress regimes, and extensional bin includes upper plates under extension and transtension. The value of ‘p’ is the probability that the distributions are statistically different based on a signed-rank nonparametric test.

53

Figure 15: Distributions of silicic calderas (left column) and other large volcanic features (right column) behind the volcanic front. (A) includes volcanic features in all types of arc settings, whereas the other rows include only those features located in the specified arc type. P-value is probability that the two distributions are statistically different based on a signed-rank nonparametric test. The distribution of silicic calderas in continental arcs is significantly different from that of more typical arc volcanoes, with silicic calderas spread broadly behind the volcanic front, whereas most other arc volcanoes occur within 10 km of the front.

54

Figure 16: Plot of caldera distances from the volcanic front versus slab dip at 100-km depth. There appears to be no correlation.

55

Figure 17: Role of pre-existing structures in silicic caldera formation. (A) Association of silicic calderas with pre-existing structures divided by arc type. (B) Comparison of distances behind the volcanic front for calderas associated with structures (top) and calderas not associated with pre-existing structural features (bottom).

56

3 Tectonic settings of silicic calderas: analysis of a global compilation

Gail Mahood was the primary advisor for this chapter. Michael Cardiff assisted with programming the naïve Bayes classifier, which is available through the electronically submitted Supplemental Materials.

3.1 Abstract

Silicic calderas are the surface expressions of large magma chambers within the upper crust, and occur in a wide variety of tectonic settings, from oceanic volcanic arcs to rifting continents. In order to understand how silicic calderas vary between different tectonic settings, we undertake an empirical, global investigation of

Quaternary silicic calderas. In particular, we focus on the relationships between caldera diameters, the composition of caldera-forming eruptions (CFE), and crustal- tectonic setting. This project incorporates data from a vast literature review and pre- existing datasets. Tectonic, compositional, and physical data were compiled for all known Quaternary silicic (trachytic, dacitic, and rhyolitic) calderas greater than 5 km in diameter, resulting in a dataset of 140 calderas.

The findings from the global compilation are as follows. The vast majority of silicic calderas meeting our criteria occur in arc settings (77%), with the remainder related to rifting, hotspots, or post-orogenic deformation. Most silicic calderas occur on continental crust, although it is important to note that ~20% of all silicic calderas are underlain by oceanic crust. Silicic calderas of a variety of compositions are

57 present on all crust types (oceanic, transitional, and continental), but the vast majority of calderas lying on continental crust are rhyolitic. The largest calderas are all in continental settings under local extension, whereas oceanic crust hosts calderas no larger than ~16 km. Relationships between diameter, silica content and the duration of pre-caldera volcanism were also examined, revealing that longer durations are associated with higher silica contents and smaller diameters. CFEs in both oceanic and continental rifting environments tend to be alkaline, consisting of trachyte or peralkaline rhyolite, whereas CFEs in arc-related calderas are generally less alkaline.

Lastly, a commonly used tectonic discrimination diagram based only on Rb, Nb and Y abundances correctly categorizes the tectonic setting of ~80% of the dataset for which trace element data were available.

Based on the accumulated data, we construct a naïve Bayes classifier that seeks to infer the regional parameters (tectonic setting, stress regime, and type of underlying crust) of a given caldera based on caldera-specific parameters (diameter, major and trace element geochemistry, and isotopic chemistry). The classification algorithm was trained with the portion of our dataset for which at least diameter and major element geochemistry were known. For those database calderas for which trace element geochemistry is available, the algorithm correctly classified 95% of the tectonic settings (volcanic arc, intracontinental, or oceanic hotspot), 70% of stress settings

(extensional or neutral/compressional), and 72% of crustal settings (oceanic, transitional, continental). Tests on the classifier reveal that trace element data was the most effective at determining tectonic setting, whereas diameter and composition

58 parameters were most effective at determining stress state. Both types of geochemical data were independently equally as effective at determining crustal setting.

In addition we used the classifier to analyze six calderas not included in the original database because of age or size constraints, in order to test whether the classifier could be used to categorize the crustal-tectonic setting of ancient calderas and possibly plutons. A significant conclusion is that although certain tectonic, crustal, and stress settings are more likely to host calderas of specific compositions and sizes, there are few true “rules” that allow unequivocal assignment of a crustal- tectonic setting to a given caldera, based solely on composition and diameter. This conclusion reflects the complexity of petrogenetic processes and regional tectonic history.

3.2 Introduction

Silicic calderas and their associated ignimbrites are the surface manifestations of large silicic magma chambers within the upper crust. In order to understand the geographic distribution of caldera-forming eruptions (CFEs) over the planet, the central issue of what causes this type of magma chamber to form in a given location needs to be researched. Based on both observation (Hughes and Mahood, 2008) and geophysical modeling (e.g., Jellinek and DePaolo, 2003; Annen et al., 2006), it is clear that a large magmatic flux into a specific region is necessary for the formation of large

(>5 km3) silicic magma chambers. Whether by subduction zone processes, crustal rifting or hotspot activity, a consistent basaltic flux from the mantle is necessary to generate a silicic magma chamber, or at least one that undergoes voluminous eruption.

What is less clear, however, is whether different tectonic and crustal settings

59 systematically yield magma chambers with specific size or compositional characteristics. Because silicic calderas are the visible results of large silicic magma chambers, they can be used as proxies for recently active magma bodies. By examining the tectonic settings, stress regimes, and underlying crustal attributes of silicic calderas worldwide, we can determine whether there is a significant link between these traits and the resulting size and composition of the calderas they host.

This study will elucidate our understanding of how regional settings focus magmatism to produce silicic centers. Throughout this paper, we use the term magma chamber loosely given the abundance of evidence suggesting that magma chambers exist in a partially or even mostly crystalline state over much of their lifetimes (e.g., Mahood,

1990; Bachmann and Bergantz, 2004).

With respect to hazard prediction, the question arises: Can any volcanic area evolve into a “super-volcano,” or does there need to be a specific confluence of tectonic or crustal factors that allows the prolonged growth of a large evolved magma chamber? Conversely, we wonder whether the measurable attributes of a given caldera, ignimbrite, or pluton can be used to determine its tectonic setting or the composition and stress regime of the enclosing crust.

This study takes a quantitative and exhaustive approach to the analysis of silicic caldera formation and its relationship to tectonic and crustal setting. We put forth a complete global database of large (> 5 km diameter), Quaternary, silicic calderas, and present our analysis of correlations between caldera-specific parameters

(caldera size, composition, and geochemistry) with regional parameters (tectonic, crustal, and stress setting) previously supposed to influence silicic magma chamber

60 formation. Based on this information we present a statistical classifier that predicts the regional parameters of calderas based on compiled caldera-specific parameters. This classification technique represents an improvement in accuracy over a commonly used tectonic discrimination diagram in assigning CFEs to tectonic environments, and we find that specific tectonic and crustal settings tend to host calderas of a certain size or compositional range. We also conclude, however, that there are few infallible rules when it comes to silicic caldera characteristics because simplistic rubrics cannot always capture the complex interaction of local and regional factors and the prior history of the crust that hosts a particular caldera.

3.3 Previous Work

Several researchers over the past six decades have discussed relationships between the characteristics of calderas and their regional settings. Smith (1979), for example, suggested that the majority of smaller calderas (e.g., Crater Lake, Oregon) form in arc settings, whereas larger, ring-fracture-bounded calderas (e.g. Long Valley,

California) form in intracontinental settings.

In addition, it has long been noted that intra-plate extensional regimes, such as the North American Basin and Range Province and the Ethiopian Rift, tend to host silicic calderas that have highly alkaline compositions, i.e., trachyte or peralkaline rhyolite. Although there is to agreement on the precise mechanisms for the generation of large volumes of peralkaline magmas, models generally attribute their formation to extensive fractionation of alkalic basalts or to partial melting induced by intrusion of mantle-derived basalt into either comagmatic alkali gabbros or refractory continental crust (Lowenstern and Mahood, 1991; Scaillet and Macdonald, 2003). In any case, we

61 predict that silicic calderas in within-plate extensional settings should have compositions that fall in the upper portion of the total-alkalis versus silica (TAS) diagram.

Tectonic discrimination diagrams were developed by Pearce et al. (1984) to assign granitic plutons to various tectonic settings based on their trace element compositions. We direct readers to this reference for a full description of the petrologic reasoning behind such categorization schemes. Based on this petrologic reasoning and analytical data from granitic plutons for which the tectonic setting is known (or inferred), Pearce (1984) developed the tectonic discrimination diagram based on Rb, Nb, and Y, which is divided into four quadrants: (1) volcanic arc granites; (2) collision-related granites (e.g., Himalayan); (3) within-plate granites (e.g.,

East African Rift); and (4) ocean-ridge granites. These diagrams have been widely used to make inferences about the tectonic setting of both plutonic and volcanic rocks

(e.g., Kay et al., 1989; Förster et al., 1997), although, as these and other studies have pointed out, the discrimination diagram has distinct limitations.

Hildreth (1981) proposed that an extensional stress regime coupled with basaltic flux from the asthenosphere favors the growth of silicic magma chambers and associated calderas. Jellinek and DePaolo (2003) affirmed this idea through geophysical modeling, noting that extension actually favors magma storage over volcanic eruption. One would thus expect to see more and larger silicic calderas in regions under local extension.

As discussed in Hughes and Mahood (2008), the nature of the underlying crust in volcanic arc settings has an impact on both caldera size and composition. Arcs

62 located on old, thick, continental crust tend to host calderas with rhyolitic CFEs, whereas CFEs in oceanic volcanic arcs tend to be dacitic. Likewise, calderas larger than 15 km in diameter tend to be located in continental arcs, and calderas in oceanic arcs to be smaller than 10 km. We examine these parameters for calderas in all tectonic settings to determine whether these generalizations derived from arc settings continue to hold.

Finally, geophysical studies indicate that the generation of significant volumes of silicic magma requires consistent intrusion of mantle-derived basalt for on the order of 105- 106 years (Annen et al., 2006). “Preheating” of the crust by repeated intrusions has been proposed to result in a change in the material properties of the wallrock, leading to a viscoelastic region where magma can be stored rather than erupting immediately (Jellinek and DePaolo, 2003). Additionally, study of Hekla in Iceland reveals that longer repose times between eruptions result in more silicic compositions

(Smith, 1979; Sigmarsson et al., 1992). Based on such studies, we hypothesize that a longer duration of pre-caldera volcanism would be associated with larger, more silicic

CFEs.

Based on this body of previous research, we identified and compiled caldera- specific parameters (caldera diameter, and composition of the CFE) and regional parameters (tectonic setting, crustal attributes, and stress regime) as well as duration of pre-caldera volcanism for each caldera.

3.4 Data

This study expands on previous work on calderas in arc settings presented in

Hughes and Mahood (2008) and Chapter 1. In the present study we examined more

63 than 800 references in order to identify calderas that meet our criteria and then collected available regional and caldera-specific data for the selected set of 140 calderas. The full reference list for the selected calderas is included in Appendix C.

An initial list of calderas with basic information such as diameter and location was compiled from previous compilations, including Newhall and Dzurisin (1988), Geyer and Martì (2008), and the Smithsonian Global Volcanism Program Database (Siebert and Simkin, 2002).

3.4.1 Caldera-specific Parameters

We set a minimum diameter of 5 km in order to eliminate incorrectly categorized sector collapses and craters. Given that caldera diameter should reflect the size of the underlying magma chamber, accidentally including these smaller features unrelated to collapse might mask magma chamber processes. Each caldera is assigned an average diameter based on the maximum and minimum dimensions of the structural caldera margin, when possible, although we recognize that the topographic margin is frequently reported in the literature.

Ignimbrite composition is considered in two different ways. In the first, the most silicic end-member of each ignimbrite is assigned a composition of trachyte, dacite (including trachydacite), rhyodacite, rhyolite (including alkali rhyolite), or peralkaline rhyolite, based on the original categorization used in the cited reference2.

We do this for the practical reason that full major element datasets are available only

2 The exception was rhyodacite, which is often lumped with rhyolite by the original authors. Those CFEs listed as rhyolitic but with silica contents of 68-72 wt% we renamed as rhyodacite.

64 for a subset of the studied calderas. For those calderas that have published chemical analyses of the caldera-forming ignimbrite, we also look at composition in terms of silica content and alkali content on a total-alkalis-versus-silica (TAS) diagram

(modified to resolve different types of silicic rocks). In addition to major element data, we examine trace elements commonly used in tectonic discrimination diagrams

(Pearce, 1984), specifically Rb, Nb, and Y. For classification purposes we compiled

87 86 additional geochemical indicators (TiO2, Zr, Sr, Ba and Sr/ Sr) that are thought to reflect subduction processes and the age and composition of any continental crustal component in the magma.

As in our previous work on volcanic arc calderas, we limit the included calderas to those that have produced CFEs within the past 2 Ma, thus ensuring that the magmatic system is in approximately the same crustal-tectonic setting as it was at its inception. In this study we record both the date of the first CFE for each caldera and the oldest age of pre-caldera volcanism. The pre-caldera duration is then the difference between the age of the earliest recorded volcanism and the age of the first CFE. This metric is likely affected by sampling bias because the oldest erupted lavas may be poorly exposed or overlain by younger deposits. The pre-caldera duration thus represents the minimum amount of time known to have elapsed between initial volcanism and caldera-formation. Calderas that are poorly studied may not have any age data for pre-caldera lavas, or even for caldera-forming ignimbrites.

65

3.4.2 Regional Parameters

Calderas were divided into three tectonic categories: volcanic arc, oceanic hotspot, and intracontinental, which we use to describe all calderas located on continental crust that are not directly related to present arc activity. Crustal type was based on regional basement descriptions from the literature, and was likewise divided into three groups: (1) oceanic, which is dominantly mafic in composition; (2) transitional, which has some continental characteristics and is at least Tertiary in age

(e.g., southern Central America); and (3) continental, which has an overall intermediate composition and is at least Mesozoic in age. Crustal thicknesses were taken largely from geophysical and regional studies.

The local stress regime surrounding each caldera was determined from published field observations, geological maps, and the World Stress Map (Heidbach et al., 2008). It should be noted that the local stress regime of an individual caldera might be substantially different from the regional stress regime. For example, a caldera in what is considered a dominantly compressional arc setting may locally experience extension or transtension because of collapse of over-thickened crust or fault interaction.

3.4.3 Results of Data Compilation

3.4.3.1 Overview

We illustrate the major findings of this compilation in Figure 19 through

Figure 25, and in the following sections discuss how these results agree or conflict with previous theories.

66

Figure 18 summarizes the locations, compositions, and sizes of the 140 calderas used in this study. Almost half the examined calderas are located on continental crust, 30% are on transitional crust, and 20% are on oceanic crust. The majority of examined calderas (about 75%) occur in volcanic arcs, leaving only 7 calderas in oceanic hotspot settings (Iceland and the Azores) and 24 in intracontinental settings (continental rifts, hotspots, or collisional zones). In addition, almost 75% of calderas occur in regions of local extension.

3.4.3.2 Diameter

As seen in Figure 19A, the quantity of calderas in each diameter bin decreases with increasing diameter. In fact, almost half of the examined calderas have diameters of less than 10 km.

3.4.3.2.1 Tectonic setting (Figure 20A)

Volcanic arc calderas and intracontinental calderas have similar diameter distributions, although it should be noted that a greater percent of the volcanic arc calderas are larger than 10 km, possibly disputing Smith’s 1979 assertion that large calderas are limited to intracontinental settings. Calderas associated with oceanic hotspots tend to be small, with the majority falling within the 5-10 km range, although the small sample size (n = 7) for this setting should be noted.

3.4.3.2.2 Crustal type (Figure 20B) and crustal thickness (Figure 20C)

The vast majority (75%) of calderas on oceanic crust are smaller than 10 km in diameter, and there are no calderas in oceanic settings larger than 16 km. Conversely,

67 on both transitional and continental crust, more than half the calderas are larger than

10 km. It is notable that calderas larger than 30 km in diameter occur exclusively on continental crust, though not exclusively in intracontinental settings. These results agree with previous observations of calderas in volcanic arcs (Hughes and Mahood,

2008), suggesting that either more silicic magma is generated in continental settings, or that continental crust is better able to store larger volumes of magma than is oceanic crust. Both of these explanations seem plausible. Continental crust tends to be thick and thus allows magma to interact with the crust and to fractionate during its longer ascent. In addition, continental crust tends to be more silicic and thus both acts as a density filter and provides a source of easily fusible material that can be assimilated.

The distribution of diameters broken down by crustal thickness (Figure 20C) does not yield any clear generalizations, but it is notable that crust less than 30-km thick can host very large calderas.

3.4.3.2.3 Stress regime (Figure 20D)

Only in extensional settings do we see calderas larger than 20 km in diameter.

Calderas located in compressional settings, which are all located in volcanic arcs, have diameters of less than 16 km. This finding supports assertions by Hildreth (1981) and

Jellinek and DePaolo (2003) that extensional settings favor magma storage in the upper crust, resulting in the development of larger magma chambers.

3.4.3.3 Composition

Approximately half of the calderas examined in this study produced CFEs that were categorized in the literature as rhyolite or peralkaline rhyolite (Figure 19B).

68

These terms reflect reporting in the literature rather than strict TAS geochemical classifications, which will be discussed in detail in the final section of the chapter.

3.4.3.3.1 Tectonic setting (Figure 21A)

CFEs associated with volcanic arcs are about equally divided between dacite, rhyodacite, or rhyolite. None are trachyte or peralkaline rhyolite (with the exception of La Primavera, Mexico, where a weakly peralkaline rhyolite erupted from a caldera in a locally extensional setting). Conversely, calderas associated with oceanic hotspots or in intracontinental settings are associated with CFEs that are trachyte, rhyolite, and/or peralkaline rhyolite.

3.4.3.3.2 Crust type (Figure 21B)

Calderas on oceanic crust tend to have CFEs that are dacitic, whereas CFEs for continental calderas are mostly rhyolitic, though it should be noted that all compositions occur in both settings. Transitional crust hosts almost exclusively the calc-alkaline sequence dacite—rhyodacite—rhyolite because volcanism in these settings is primarily related to arc magmatism (such as southern Central America and northern Japan).

3.4.3.3.3 Crustal thickness (Figure 21C)

Almost 75% of silicic calderas located on crust over 40-km thick are rhyolitic, whereas CFEs on crust less than 40-km thick more evenly occupy the entire compositional range. This observation, combined with the fact that continental crust tends to host both large and rhyolitic calderas, suggests that the generally greater

69 thickness of continental crust contributes to the concentration of large, rhyolitic calderas in continental settings. We note, however, that the Taupo Volcanic Zone,

New Zealand, is located in a region of thin continental crust yet hosts several large rhyolitic calderas. Such variations suggest that exceptions to “rules” may provide valuable insights into the complexities of magma genesis and storage. The Taupo

Volcanic Zone, for example, is located in a presently unusual tectonic setting where a continental arc is undergoing strong back-arc transtension.

3.4.3.3.4 Stress regime (Figure 21D)

Caldera-forming eruptions in compressional settings are exclusively in the calc-alkaline range (dacite—rhyodacite—rhyolite) because all compressional settings are located in volcanic arcs. Calderas in extensional settings fall into all compositional categories but the majority are rhyolite or peralkaline rhyolite, again affirming the notion that extensional settings favor magma storage and fractionation.

3.4.4 Relationship between Diameter and Composition

Because compositions in the literature are not necessarily uniformly defined, we also examined composition in terms of highest CFE silica content, which was available in the literature for ~90% of the examined calderas. In order to determine whether more silicic CFEs are associated with larger calderas, we plot caldera diameter versus silica content (Figure 22). Calderas with diameters less than ~20 km have CFE silica contents that range over almost the entire compositional spectrum. In contrast, calderas larger than 20 km all have CFE silica contents greater than 72 wt%

SiO2. This figure also reaffirms the observations from Figure 20B and D: There is a

70 wide range of silica contents for all crustal types and stress regimes, but calderas larger than 20 km are all located in extensional, continental settings.

3.4.5 Duration of Volcanic Activity

Figure 23 shows the relationship between the minimum duration of pre-caldera volcanic activity and both CFE highest silica content (Figure 23A) and caldera diameter (Figure 23B). Longer durations of pre-caldera volcanism are associated with higher silica contents. The converse, however, is not true; there is a wide range of silica contents for short pre-caldera durations. Perhaps counter-intuitively, caldera diameter (a proxy for the related variable CFE volume) does not increase with duration of pre-caldera volcanism. In fact, the largest calderas (>30 km), all of which associated with rhyolitic CFEs, appear to be associated with durations of pre-CFE volcanism of less than two million years.

The observation that extended periods of pre-caldera volcanism result in rhyolitic CFEs is logical if the longer a magma chamber exists, the more it will continue to fractionate, assimilate material, and evolve. The observation that larger calderas are associated with shorter pre-caldera durations of volcanism is more difficult to explain. One possible explanation is that a magma chamber that is tapped later in its solidification history will have highly differentiated but smaller volumes of eruptible magma. Alternatively, there may be a sampling bias for very large calderas in that their voluminous eruptive products may literally cover-up the pre-caldera history of an area.

71

3.4.6 TAS Diagrams

Major element data were available the CFE for 117 of the 140 examined calderas. Major elements were normalized to a 100% total, eliminating trace elements and volatiles. Using a modified TAS diagram that partitions the silicic compositions more finely than the standard version of LeMaitre (1989), we illustrate in Figure 24 the compositions present in various regional settings. The CFEs in volcanic arcs are typically characterized in the literature as dacite, rhyodacite, or rhyolite, but Figure

24A demonstrates that a significant fraction of them are more alkalic, falling in the trachydacite or alkali rhyolite fields. In this plot and Figure 24B, for intraplate settings, one can see some differences in CFE composition based on crustal type and tectonic setting. Oceanic arc calderas tend to fall within the dacite and rhyodacite fields, whereas continental margin calderas tend to be skewed towards the rhyolitic end of the plot. Calderas in intraplate settings are far more alkalic as a group, mostly falling in the alkali rhyolite and trachyte fields. In contrast to arc settings, the entire silica range is more or less equally represented in CFEs from oceanic hotspots and intracontinental calderas. As noted previously in the discussion of Figure 20, the largest calderas tend to be rhyolitic and located in continental margin arc and intracontinental tectonic settings.

3.4.7 Discrimination Diagrams

Of the 140 examined calderas, 67 had available trace element data for the

CFEs. We use these 67 systems to test the effectiveness of the tectonic discrimination diagram based on Rb, Y, and Nb developed by Pearce et al. (1984) for use with

72 granites. Most of the calderas in volcanic arcs fall within (or very near) the volcanic arc quadrant of the diagram (Figure 25A). The exceptions are a few large continental calderas that fall squarely in the within-plate granite field: La Primavera, Mexico, Pino

Hachado in the Southern Andes, and Kapenga of the Taupo Volcanic Zone. It is notable that all three of these calderas lie in continental arcs that are locally experiencing extension or transtension. Likewise, most of the intracontinental and hotspot-related calderas fall in the within-plate granite quadrant (Figure 25B), except

Long Valley, California, which plots squarely in the volcanic arc quadrant. Iceland and the Azores plot in the within-plate field, presumably because they reflect locales where hot-spot magmatism coincides with a spreading ridge. If we allow Iceland and the Azores to be considered “within- plate,” although they sit on mid-ocean ridges, this discrimination diagram correctly assigns the tectonic setting of about 80% of the

67 calderas.

It is not surprising that none of the modern calderas plot in the ocean-ridge quadrant, given that typical mid-ocean ridges and back-arc spreading centers lie in deep water and so would not produce readily visible calderas or ignimbrites. It is possible that there are small silicic calderas in these settings that have not been detected given the resolution of currently available bathymetry, but given the low fluxes of basalt and the absence of continental crust, generation of large volumes of silicic magmas would not be expected (Dixon-Spulber and Rutherford, 1983). The lack of calderas in the syn-collisional quadrant may likewise be due to their relatively small volumes, resulting from their genesis dominated by crustal melting (e.g., Patino

Douce, 1999). In addition syn-collisional magmas such as the Himalayan

73 leucogranites tend to be “wet” and solidify during their ascent (Cann, 1970), making association with CFEs highly unlikely.

3.5 A New Bayesian Classification Scheme

Being able to determine the regional parameters for silicic calderas, plutons, or ignimbrites at the time of their formation would be useful for studying ancient geological terranes. As discussed, discrimination diagrams alone have frequently been used in this capacity. In order to examine automated methods for estimating regional parameters based on available data, we utilize a naïve Bayes’ classifier. In a Bayes’ classifier, we assume that each class (i.e., the regional setting parameter) leads to a particular distribution for each of the measured features (i.e., the caldera-specific parameters). Given a set of possible classes (C1,C2,…,Cm) and a set of features

(F1,F2,…,Fn), we seek to find the most probable class by performing the following optimization:

(1) max p(Ci | F1,..., Fn ) ! p(Ci ) p(F1,..., Fn | Ci ) Ci which corresponds to finding the class with the maximum a-posteriori probability, as calculated using Bayes’ theorem, where p(Ci) is the prior probability of each class and p(F1,…Fn|Ci) is the likelihood of a particular feature set given the class.

The naïve feature of the naïve Bayes’ classifier is that it makes the assumption that the feature distributions are independent, such that the optimization problem can be re-written as follows:

n

(2) max p(Ci | F1,...Fn ) " p(Ci ) p(Fj | Ci ) C ! i j=1

74

In practice, each of the distributions p(Fj|Ci) are then estimated by assuming a particular form for the distribution (e.g., normal or lognormal), and then determining the relevant parameters of those distributions (e.g., mean and variance) by examining available data for which the classes are known. We refer to this as “training” of the naïve Bayes’ classifier. Once the naïve Bayes’ classifier is trained, it can be used to estimate the unknown class of a particular caldera given only its features.

Given the previous discussion, the “naïve” assumption that all feature data are independent is obviously not true; indeed, we expect that many features have correlated distributions. Still, many studies (e.g., Rish, 2001) have noted that the naïve

Bayes’ classifier works quite well even when the naïve assumption is violated. This is primarily because it does not matter whether the probability of each class is calculated exactly correctly; it only matters that the correct class has the highest probability. All the features considered in the classification scheme are listed in Table 4, where each feature is assigned an assumed distribution (normal or log-normal) based on a review of feature histograms. Some features, such as diameter, have an obviously exponential distribution but we find that assuming a normal or log-normal distribution often leads to more accurate classification. This is probably because the extreme values hold considerable information, and the normal distributions tend to give more weight to extreme values.

3.5.1 Results

Once trained on caldera data that included at least diameter, composition, and major element geochemistry, we tested the classifier on various subsets of the

75 database, summarized in Table 5, rows 1-3. , Table 5 row 3 in particular can be used to compare our classifier with the commonly used tectonic discrimination diagram.

The classifier correctly classified 95% of the tectonic settings (volcanic arc, intracontinental, or oceanic hotspot) of these 67 calderas, compared to the 80% classified correctly by the Rb versus (Nb+Y) tectonic discrimination diagram of

Pearce et al. (1984). For this subset of the data, the classifier correctly categorized

72% of crustal settings (oceanic, transitional, continental), and 70% of stress regimes

(compressional, transpressional, or neutral versus transtensional or extensional).

For the 117 calderas for which there are, at a minimum, data for the three major-element parameters from Table 4, the classifier correctly categorizes nearly all

(94%) of the calderas with respect to tectonic setting. It is less effective (66%) at correctly categorizing crustal type into continental, oceanic, or transitional. We are most concerned, however, with miscategorization of oceanic calderas as continental and vice versa. The algorithm only miscategorizes 10% of the crustal settings in this manner, suggesting that the algorithm can differentiate between oceanic and continental settings. It is only somewhat better than chance (64%) at assigning calderas to the two classes of stress regime. This is probably due to stress regime being difficult to characterize because although the structural evolution of some caldera-bearing regions has been studied in careful detail, many have been only marginally examined and the stress regime at the time of caldera formation is poorly constrained. Even in well-studied areas, caldera magmatism may be fairly short-lived and the tectonic setting that is assigned may be either current or time-integrated based on regional patterns of faulting, which may not apply at the time of volcanism.

76

In order to determine which caldera-specific parameters were most effective in determining the regional classifications, we ran three tests in which different parameters were eliminated from the training dataset and then observed how many calderas the algorithm categorized correctly (Table 5 rows 4-6). Only calderas with at least Rb, Y, and Nb were used in the tests. We find that trace element data were the most effective at determining tectonic setting (row 3), whereas diameter and bulk composition parameters were most effective at determining stress state (row 1). Both types of geochemical data were independently equally as effective at determining crustal setting. Using only the composition code and diameter in the training (row 2) resulted in basically a random assignment of calderas to regional classifications.

In addition to testing the classifier with the data used in its construction, we applied it to several calderas not included in the original study because of size or age constraints (Table 6 and Figure 26). These included regional settings (syn-collisional, slab “break-off” or abrupt steepening) and compositions (strongly peraluminous) not represented in the original, modern database. Three of the calderas—Huichapan,

Negra Muerta, and Caetano—are older than 2 Ma, but are young enough to be in tectonic settings that are fairly well understood. The classifier successfully assigns

Huichipan, Mexico to its continental arc setting under extension. The classifier assigns the same regional setting to Negra Muerta and Caetano calderas, which are thought to have formed in response to abrupt steepening or break off of a subducting slab following an amagmatic interval of shallow subduction (Humphreys, 1995; Kay and Coira, 2009). This regional setting was not one represented by the calderas in our database, so it is not available as an assignment category. Presumably the classifier

77 assigns these two calderas to an arc setting because the magmas contain a crustal component that has an arc geochemical signature.

One of the calderas tested—Socorro—is smaller than the 5-km cutoff for our database. The algorithm classifies all the test calderas as being in continental settings, but Socorro caldera is located in an oceanic setting on the archipelago of Islas

Revillagigedo, which coincides in part with an abandoned spreading center near the mouth of the Gulf of California. Socorro was classified as being intracontinental, but given the geochemical similarities between calderas located in hotspot-spreading ridge settings and intracontinental settings (Figure 24B and Figure 25B), this miscategorization is not surprising.

Finally we apply the classifier to two thick intracaldera ignimbrites preserved as roof pendants in otherwise largely plutonic terrains: the Cretaceous Minarets caldera in the Sierra Nevada batholith (Fiske and Tobisch, 1994; Lowe, 1996) and the

Late Devonian Violet Town Ignimbrite of the Lachlan Fold Belt of Australia. The latter is unusual among volcanic rocks for its strongly peraluminous composition, which ties it petrogenetically to nearby S-type granites (Clemens and Wall, 1984;

Kemp et al. 2008). It classifies them both as continental, extensional, and volcanic- arc-related. This is consistent with the well-accepted continental volcanic arc setting of the Minarets caldera, though the exact stress regime at the time remains under study. The assignment of the Violet Town Ignimbrite is in agreement with the suggestion by Kemp et al. (2008) of a continental-margin arc for the tectonic setting of the Lachlan Fold belt, but is at odds with Clemens and Wall (1984) argument for an intracontinental post- or anorogenic setting.

78

3.6 Discussion

Based on our compilation, it is possible to make generalizations about the occurrence and characteristics of silicic calderas in different crustal and tectonic settings. There are, however, few true “rules.” Our findings thus emphasize the complexity of magma formation; tectonic setting, the nature of the underlying crust, and the duration of pre-caldera volcanism all have an apparent effect on the size and composition of silicic calderas. But using the diameter, composition, and geochemical signature of a given caldera, ignimbrite or pluton to nail down its precise tectonic- crustal setting is, at best, a probabilistic endeavor. Magma trace element geochemistry, for example, can be affected by tectonic events prior to the formation of the magmatic system. In western North America, for instance, shallow subduction of the Farallon plate during the Laramide likely added an arc signature to the lithospheric mantle at distances exceeding 2000 km from the trench (Lee, 2005), which became incorporated in magma erupted during the subsequent mid-Tertiary “ignimbrite flare- up” and during volcanism associated with later Miocene extension. The greater the proportion of crustal melt in the silicic magma, the more its trace-element composition will reflect the nature of that crust instead of the tectonic setting in which the magma formed.

The naïve Bayes classifier combines a great deal of data to probabilistically classify calderas based on regional parameters. This method shows significant potential in correctly classifying the tectonic and crustal settings of calderas, being more successful than the classic Rb versus Nb + Y discrimination diagram, although it does poorly at classifying calderas based on stress regime. One problem with widely

79 implementing such a model is that it is “tuned” to the current global distribution of silicic calderas. Ideally additional data could be added to expand the model, and we suggest that the use of a naïve Bayes classifier in examining ancient ignimbrites or plutons may represent an area of future fruitful research. In particular, to make a classifier more widely applicable, we would need to include data for older calderas that occurred in tectonic settings not represented in our database of modern calderas.

One example is the calderas that follow intervals of flat-slab subduction (e.g., the voluminous ignimbrites of the mid-Tertiary of the Basin and Range and of the Mio-

Pliocene of the Altiplano) that are thought to be a result of asthenospheric mantle upwelling in response to slab break-off, abrupt steepening of the slab angle, or lithospheric delamination. Other classes of settings are syn-collisional and post- collisional, which, in practice, will be difficult to separate because the former segues into the latter, and also occurs at different times along an arc. It might however be possible to distinguish centers in post-collisional rifts from those in transpressive zones.

3.7 Conclusions

We compile and review regional and caldera-specific parameters for 140 large, silicic, Quaternary calderas worldwide. Our goal was to discern whether there are consistent differences between silicic calderas in different regional settings in terms of size, composition and isotopic geochemistry. In addition, we develop a naïve Bayes classifier that attempts to categorize the regional parameters of calderas (tectonic setting, nature of the underlying crust, and stress regime) based solely on caldera- specific parameters. We find that:

80

1) Silicic calderas larger than 10 km in diameter occur in all crustal settings—

even in ocean island arcs such as the Izu-Bonin arc.

2) Calderas in volcanic arcs and intracontinental settings have similar size

distributions, at least for those features over 5 km in diameter. Calderas larger

than 20 km in diameter occur almost exclusively in continental, extensional

environments, regardless of tectonic setting.

3) A wide variety of CFE compositions can develop in many crustal and tectonic

settings although volcanic arcs generally host dacitic, rhyodacitic, and rhyolitic

CFEs, whereas intracontinental and oceanic hotpot settings host more alkaline

compositions.

4) Although calderas associated with oceanic crust tend to be less silicic than

those on transitional or continental crust, rhyolitic calderas do occur in oceanic

settings and, conversely, dacitic calderas can be found in continental settings,

even where the crust is thicker than 40 km.

5) The duration of pre-caldera volcanism shows a slight positive correlation with

CFE silica content, whereas caldera diameter does not.

6) A naïve Bayes classifier correctly classifies the tectonic and crustal settings for

most of the calderas in the database but is less successful at classifying stress

regime. In addition, the classifier had some difficulty classifying the tectonic

settings of some calderas not originally included in the database.

7) Examination of the compilation results and the successes and failures of the

naïve Bayes classifier shows that although calderas of certain sizes and

ignimbrite composition tend to occur in different regional settings, there are no

81

true rules that would allow unambiguous assignment of tectonic setting, crustal

type or stress regime to ancient calderas and to silicic plutons.

8) The Bayesian classifier and other tectonic discrimination diagrams works best

in "predicting" modern tectonic settings because it is capturing the composition

of both the mantle and crustal components contributing to the arc magma: arc

basalts and continental lithosphere shot through with arc component. It is a

little less effective predicting type of crust, presumably due to the range of

crustal versus asthenospheric mantle component present in the silicic CFE.

The fact that it is poor at predicting stress state suggest either that stress state

has only a weak control on the development of silicic magma chambers or that

the local structures and stress state that affect magma chamber formation may

not be expressed in the larger-scale stress state we compiled in the data base.

Such data are integrated over space and time. We now know from high-

resolution dating of silicic systems that such bodies can accumulate in periods

of tens to hundreds of thousands of years. This creates the possibility that

sizable silicic magma chambers form during transient conditions that favor

magma accumulation and may differ from the tectonic data we compile in

which the setting is integrated over space and time.

82

3.8 Tables

TABLE 4: MODEL PARAMETERS INCLUDED IN BAYESIAN CLASSIFIER % calderas for Assumed Parameter Significance which parameter distribution is known 1 Average diameter (km) Figure 20 - Calderas larger than 20 normal 100% km in diameter do not occur in oceanic settings or in compressional regimes

Composition code Figure 21- Dacite, rhyodacite, and normal 100% rhyolite dominate in arc settings, whereas trachyte and alkali rhyolite predominate in intraplate settings.

Silica content Figure 24 - Higher silica contents normal 91% are associated with transitional to continental crust.

Total alkalis Figure 24 – Higher alkali contents normal 84%

(Na2O + K2O) are associated with intracontinental and oceanic hotspot settings.

TiO2 Tends to be high in normal 84% intracontinental settings1.

Rb Figure 25– Higher Rb content with log normal 61% respect to Nb + Y is associated with intracontinental or oceanic hotspot settings. Lower values associated with volcanic arcs2.

Nb + Y Figure 25– Higher (Nb + Y) log normal 48% associated with intracontinental or oceanic hotspot settings. Lower values associated with volcanic arcs2.

Zr Tends to be high in log normal 61% intracontinental settings2.

Ba High Ba/Nb reflects larger input log normal 59% from a subduction-enriched component2.

Sr Removed during the fractionation log normal 62% of plagioclase.

87Sr/86Sr isotopic ratio Isotopic values higher than 0.706 normal 44% associated with cratonic continental crust.

Notes: 1 – The distribution is exponential, but using a normal distribution in the model results in more correct classifications; 2 – From Pearce et al. (1984). Table 4: Model parameters included in the naive Bayesian classifier

83

TABLE 5: EVALUATION OF NAÏVE BAYES CLASSIFIER

Percent correctly Percent correctly Percent correctly classified: Tectonic classified: Crustal classified: Stress setting Parameters setting regime Calderas categorized included (volcanic arc, oceanic (oceanic, transitional, (compressional/neutral, hotspot on MOR, continental) extensional) intracontinental)

All All (140) 64% 64% 87%

Calderas with at least major element All 66% 65% 94% geochemistry data (117)

Calderas with at least major element All 72% 70% 95% and Rb, Y, and Nb data (67)

Only diameter Calderas with at and major least major element 62% 67% 78% element and Rb, Y, and Nb parameters data (67)

Only Calderas with at composition least major element 38% 64% 37% code and and Rb, Y, and Nb diameter data (67)

Only trace Calderas with at element and least major element 62% 53% 91% isotope and Rb, Y, and Nb parameters data (67) Notes: All use model trained with subset of calderas that have major element geochemistry available, in addition to caldera diameter. Number in parentheses in column two is the number of calderas categorized. Caldera data include in Appendix B, with references in Appendix C.

Table 5: Evaluation of naïve Bayes classifier

84

- rently volcanic arc volcanic arc volcanic arc volcanic arc volcanic arc Classification intracontinent tectonic setting

indicates an appa extensional extensional extensional extensional extensional extensional Classification stress regime

Bohrson and Reid (1997), 7

-

crust Greene and Schweikert (1995). Full continental - continental continental continental continental continental Classification

John et al. (2008), 6 - 5 dacite rect classification, whereas normal rhyolite rhyolite rhyolite trachyte rhyodacite peralkaline Composition (1995),

5 15 27 21 8.5 9.5 Fiske and Tobisch (1994), 11 indicates a cor - Diameter d Humphreys - 4 bol

-

– -

-

(2009), margin arc - continental - Lowe (1995), 10 - subduction - unknown unknown stress regime under extension stress regime compressional tectonic setting – oceanic hotspot Post plate extension? ntracontinental or Apparent crustal I continental; within arc Abandoned oceanic spreading ridge and Continental volcanic Sierra Nevada Arc Orogenic continental Kay and Coira - TABLE 6: CLASSIFICATION OF TEST CALDERAS AND IGNIMBRITES

Range, Kemp et al. (2008), 9 - Mexico Mexico Location SE Australia, Nevada, USA Volcanic Belt, Central Sierra Puna Plateau, Central Andes Socorro Island, Central Mexican Petrinovic (2005), 3 Basin and Nevada, CA, USA Lachlan Fold Belt, -

–

,11 Don – 9,10

,8 7 Verma (2001), 2 -

Clemens and Wall (1984), 8 geochemical data given in Appendix A. For the classifications, incorrect classification, italic indicates that true classification is uncertain. 1 : 1

,5 Minarets 2,3 4 6

Name, Age (1) Huichipan Guinyó Ignimbrite, 4.2 Ma (2) Negra Muerta Acay Ignimbrite, 9 Ma (3) Caetano, 34 Ma (4) Socorro, 0.37 Ma (5) Violet Town Volcanics, Late Devonian (6) Caldera, Cretaceous, Notes

Table 6: Classification of test calderas and ignimbrites

85

3.9 Figures

Figure 18: Global distribution of the 140 examined silicic calderas, coded by size, caldera-forming eruption (CFE) composition, and tectonic setting. In text, non-arc continental calderas are grouped as intracontinental.

86

Figure 19: Distributions of (A) average diameters and (B) CFE compositions for all examined calderas. The number of calderas in each diameter bin decreases with increasing diameter, and the most common CFE composition is rhyolite.

87

Figure 20: Distributions of caldera diameters differentiated by crustal and tectonic attributes. Calderas in oceanic and compressional settings are generally small. Calderas in intracontinental settings and in volcanic arcs on continental crust can exceed 30 km.

88

Figure 21: Distributions of caldera compositions, differentiated by crustal and tectonic setting. Alkaline compositions (trachyte and peralkaline rhyolite) are restricted to extensional intracontinental or oceanic hotspot settings, whereas dacitic and rhyodacitic compositions are restricted to arc settings. Crust thicker than 40 km hosts almost exclusively rhyolitic calderas.

89

Figure 22: Highest reported silica content for each examined caldera-forming eruption versus average caldera diameter. Although there are CFEs of each composition in the caldera diameter range of 5-20 km, those associated with calderas larger than ~ 25 km are all rhyolitic and in extensional, continental settings.

90

Figure 23: Duration of volcanism prior to the first CFE versus (A) highest silica content of the CFE and (B) caldera diameter. Perhaps counter-intuitively, caldera diameter (also a proxy for the related variable CFE volume) does not increase with duration of pre-caldera volcanism. However, the greater the pre- caldera duration, the more likely the CFE will be rhyolitic.

91

Figure 24: Total alkalis versus silica diagram. Although CFEs in volcanic arcs tend to be in the calc- alkaline dacite-rhyodacite-rhyolite range, there is a significant fraction that falls into the more alkalic trachydacite and alkali rhyolite fields. Trachytic CFEs occur almost exclusively in intraplate settings

92

Figure 25: Rb versus (Nb + Y) tectonic discrimination diagrams of Pearce et al. (1984) for calderas in (A) volcanic arc settings and (B) intracontinental or oceanic hotspot settings. About 80% of calderas are correctly categorized by this diagram compared to 95% using the naïve Bayes classifier.

93

Figure 26: Locations of test calderas (not included in the database) on the modified TAS and Pearce et al. (1984) discrimination diagrams. See Table 3 for caldera descriptions.

94

4 Reinvestigation of the 1989 Mammoth Mountain, California Seismic Swarm and Dike Intrusion

Data for this paper was supplied by Dr. John Langbein of the USGS, Menlo Park,

California and by Dr. Dan Dzurisin of the Cascade Volcano Observatory, Washington.

Dr. David Hill and Professors David Pollard and Gail Mahood provided helpful discussion.

4.1 Abstract

During the period of unrest and inflation beneath Long Valley, east central

California, between 1988 and 1992, a swarm of seismic activity occurred beneath

Mammoth Mountain, on the caldera rim in 1989 followed by the onset of diffuse CO2 degassing. These events, combined with elevation changes along a leveling line north of Mammoth Mountain suggested a dike intrusion. An early inversion of deformation data resulted in a dike model with an unrealistically small amount of opening (13 cm) for the proposed dimensions. The current reinvestigation of the 1989 seismic event at

Mammoth Mountain, CA incorporates relocated earthquakes (Prejean et al., 2003) and ellipsoidal inflation sources beneath the caldera to the east (Langbein et al., 2003) to better constrain the location and dimensions of a proposed dike intrusion beneath

Mammoth Mountain. Assuming that the dike coincides with a “keel” of seismicity at

7-9 km depth, most of the dislocation parameters can be constrained. Fixing all dike parameters except opening to coincide with the keel, inversion of the local leveling

95 data yields a dike with 3.3 m of opening. Allowing the dike height to vary as well improves the fit to the data significantly and results in a dike 1 m wide and nearly 7 km km high. These results indicate that the intruded dike had significantly more opening than previously modeled. Magma volumes for reasonable models center around 0.015 km3, suggesting a small, likely basaltic, intrusion.

4.2 Introduction

On May 4, 1989, seismic activity initiated beneath Mammoth Mountain, a dacitic volcano on the southwestern rim of Long Valley caldera in eastern California

(Figure 27). The swarm included approximately 2700 small (M ≤ 3.4) earthquakes

(pink and black dots, Figure 28) over a period of eleven months (Hill et al., 1990). A leveling line north of the volcano (blue dots, Figure 28), along Route 203, recorded characteristic deformation (Figure 29), which Hill et al. (1990) attributed to a small dike intrusion beneath Mammoth Mountain. The onset of diffuse CO2 emissions on the flank of the mountain following the swarm and its magmatic isotopic signature gave further evidence for an intrusion (Farrar et al, 1995). The Mammoth Mountain seismic swarm was part of a larger episode of unrest in the Long Valley caldera, which consisted of significant uplift beneath the resurgent dome as well as earthquake swarms along the south moat of the caldera. Subsequent research into the event has focused on modeling the dike intrusion in conjunction with other Long Valley deformation sources (Langbein et al., 1993; 1995; and 2003), relocating the 1989 earthquake swarm (Prejean et al., 2003), and study of the emitted CO2 (Farrar et al.,

1995). In this paper, inversion techniques are combined with the results of Prejean et

96 al. (2003) and Langbein et al. (2003) to better constrain the dimensions of the intruded dike, seeking to elucidate the nature of magmatic unrest in this part of California.

4.3 Geological Setting and Motivations

Mammoth Mountain is a dacitic cumulovolcano that was formed by multiple dome-building eruptions between 68 ky – 58 ky before present (Mahood et al., 2010).

Though it is located near Long Valley caldera, which underwent catastrophic caldera collapse at 760 ky, Mammoth Mountain is a distinct magmatic system and represents the local westward migration of focused magmatism from the center of the Long

Valley volcanic field (Hildreth, 2004). Holocene volcanism in the region includes the basaltic cinder cones of Red Cones, which formed at ~9 ky, four kilometers southwest of the mountain (Bursik, 2004), and the 600-year-old rhyolitic Inyo Domes that extend northward towards Mono Lake. Evidence for a few young phreatic eruptions are located on the north flank on Mammoth Mountain, one within a ski area, which may have been associated with the dike that fed the Inyo domes to the north (Hill et al.,

1990). These small eruptions are occurred within the Holocene, and so a future basaltic eruption is certainly possible. Developing an understanding of the Mammoth

Mountain magmatic system is thus of vital importance for determining likely hazards posed by future activity and characterizing the progression of magmatism in the Long

Valley region over time.

4.4 1989 Intrusion and Earthquake Swarm – Previous Work

Hill and Prejean (2005) present an excellent summary of previous work on

Mammoth Mountain. Deformation at Long Valley between 1988 and 1992 was

97 recorded by a combination of 2-color electronic distance meter (EDM) and leveling lines, although only the Route 203 leveling line recorded deformation at Mammoth

Mountain (Figure 29). The leveling profile across the northern flank of Mammoth

Mountain shows a W-shaped uplift pattern about 10-13 km west of the CASA leveling mark (Figure 29), near the Casa Diablo geothermal plant. Modeling of both the EDM and leveling data by Langbein et al. (2003) includes two ellipsoidal sources: one beneath the resurgent dome and one beneath the south moat (green dots, Figure 27).

Based on the Route 203 leveling data, Langbein et al. (1993 and 1995) suggested that the dike extended from a depth of ~12 km to within ~2 km of the surface over a length of 6 km, with an average opening of about 13 cm wide, although they noted that the dike dimensions were poorly constrained. Subsequent workers suggested that a dike with this aspect ratio was unlikely to be entirely magma-filled, and postulated that the upper part of the intrusion was filled with CO2 or aqueous fluids (Foulger et al., 2003).

Initial analysis of the 1989 seismicity showed a slab-like hypocenter geometry almost “collinear” with the dike that fed the Inyo Domes (Hill et al. 1990). T-axes for the fault plane solution were oriented perpendicular to the strike of the plane, consistent with a dike intrusion (Hill et al., 1990). Prejean et al. (2003) relocated the hypocenters for the 1989 seismic event using a double-difference method, revealing two seismically defined structures. The first was a “keel” 9 to 7 km beneath the surface, 2 km in length and 0.5 km wide with T-axes oriented perpendicular to its strike. The second was a shallower ring-like structure around which seismicity traveled over time. Its defining seismicity was likely associated with a small amount

98 of normal slip (1 cm) on a preexisting structure, reactivated by migrating fluids from the lower magmatic dike (Prejean et al., 2003).

By combining the geometry from Prejean et al.’s (2003) relocated earthquakes and the ellipsoidal models for Long Valley uplift (Langbein et al. 1995), I have investigated the implications of new constraints on the inversion of the Route 203 leveling data for the possible dimensions of a dike beneath Mammoth Mountain.

4.5 Methods

4.5.1 Model

The dike is modeled with an Okada-type opening dislocation (Okada, 1985) in an elastic halfspace. We take the local x, y, z coordinate system to be east, north, down, with depth being positive. The model has eight variables: length, height, depth to bottom edge (positive), strike, dip, center of bottom edge (x,y) and opening. It is assumed that the orientation (strike and dip) and horizontal position (x, y) of the dike are fixed by the “keel” observed in the relocated earthquakes (Prejean et al. 2003), and thus solve only for the dike dimensions in a series of increasingly less constrained inversions. We also assume that displacement due to ~ 1 cm of normal slip on the ring structure is negligibly small compared to the dike opening, given the dike-like displacement profile of the leveling line. Uplift at the eastern end of the leveling line due to deformation within the caldera is modeled using the two prolate spheroid sources proposed by Langbein et al. (2003) (see note in Table 7).

99

4.5.2 Trials

A series of model trials were investigated, ranging from highly constrained to weakly constrained in terms of dike dimensions. All trials consist of a vertical, rectangular dislocation beneath Mammoth Mountain, collinear with the seismic keel.

In Trial 1, dike length is fixed to 2.3 km, dike height is fixed to 2 km, and the base of the dike is set at 9 km depth, based on the locations of the relocated earthquakes, thus leaving only dike opening as an unknown. In Trials 2 and 3 I also allow dike height and dike length, respectively, to vary, maintaining the bottom of the dike at 9 km depth. Dike height can vary between 0 and 9 km, and length can vary up to 20 km northeast of the southwest endpoint of the seismic keel. Theoretically, the dike could extend in the opposite direction from the keel as well, but given the location of the leveling line, extending the dike toward the southwest does not significantly affect the modeled displacements. In Trial 4, dike opening, height, and length are allowed to vary simultaneously, again holding the base of the dike at 9 km. Finally in Trial 5, allow all four dimensional parameters (opening, height, length, and depth) are allowed to vary, where the depth and height of the dike are constrained to the top 20 km of the crust, and it is assumed that the dike does not break the surface.

Rubin (1995) suggests that for basaltic dikes, opening is typically on the order of 1 m, associated with a magmatic overpressure on the order of 1-10 MPa.

Overpressure is here defined as the difference between pressure within the dike and

the least compressive principle stress, or P − s1; s1 is assumed to be constant over the dike. Overpressure is proportional to the ratio of the dike opening w, to the dike’s minimum dimension l: €

100

w 2(P #s1)µ " (3) l (1#$) where µ is the shear modulus (assumed to be 30 GPa) and v is Poisson’s ratio. We can ! thus estimate the dike overpressure for a given geometry, which we conservatively constrain to be between 0 and 100 MPa.

4.5.3 Data

Surveys along the Route 203 leveling line were fully completed in 1988 and

1992, and the uplift over this time period is shown in Figure 29. Because of local deformation associated with geothermal energy production near CASA, the benchmarks between CASA and benchmark 34EGE were removed from the dataset, leaving a total of 26 benchmarks used in the inversion. Based on the data given in

Figure 11 b of Langbein et al. (1993), the total uplift at 34EGE is 9.64 cm, and we assign an error of 0.5 cm to this data point, based on measurement error. Uplift along the leveling line is then calculated by iteratively adding the measured displacements between benchmarks, resulting in an uplift profile (Figure 29):

(4) d = Al where, for an n-point leveling line, d is an n x 1 vector containing the uplift at each benchmark, A is a lower! triangular n x n matrix of ones, and l is an n x 1 vector containing the displacements between each successive benchmark.

Breaking with the convention laid out by Langbein et al. (1993) we have chosen to fit the total uplift at each benchmark rather than the displacements between benchmarks. This choice results in a data covariance matrix, ∑d, with off-diagonal

101 terms and increasing standard error for each uplift measurement along the leveling line from east to west:

T (5) "d = A #A where σ is the vector of standard errors for each leveling measurement.

! 4.5.4 Inversion

The inversion process is divided into two parts. In the first, a simplex search algorithm available in the Matlab optimization toolbox (Lagarias et al., 1998) is used to find a best-fitting model, starting from a number of initial guesses. We pose the inversion in this step such that dike parameters, m, are chosen to fit the data d by minimizing the L2-norm of the misfit weighted by the data covariance, i.e. assuming

Gaussian σ:

T "1 (6) min(d " G(m,x,y)) #d (d " G(m,x,y))

where x and y are the east and north coordinates of the leveling benchmarks, G is the ! model consisting of the dike and the two prolate spheroids described in Langbein et al.

(2003). The expression being minimized is also referred to as the weighted-r2 value, where r is the vector of differences between the measured and the modeled values.

Because such search algorithms can fall into local minima, we initiate the search from a series of 100 randomly generated starting points. Given the small number of free parameters (1 to 4) as compared to the number of data points (26), we find that this method converges to the best solution within 10 iterations.

To determine how certain we are of our estimate of dike opening, we have implemented a Markov Chain Monte Carlo (MCMC) simulation in order to generate a

102 posterior distribution of inverted parameter values. This technique has the distinct advantage of demonstrating the probabilistic range of feasible solutions; a complete description can be found in Mosegaard and Tarantola (1995). This method relies on a

Bayesian formulation in which we seek to sample the probability distribution of model parameters given the data, again assuming a Gaussian distribution of error:

#n / 2 #1/ 2 T #1 (7) L(m | d) = (2") (det($d )) exp(#r $d r /2) where L is the likelihood of the model given the data and n is the number of data points. !

4.6 Results (Figure 30 and Table 7)

Table 7 summarizes the results of the inversion for each trial. Forcing the dike geometry to coincide with the relocated hypocenters and solving only for dike opening

(Trial 1) results in an opening of 3.32 m and an associated overpressure of 33 MPa.

This model tends to over-predict uplift at the eastern end of the leveling line and under-predicts uplift at the western end, where the dike intrusion is assumed to have occurred. The posterior probability distribution of dike opening shows that realistic values of opening range from near zero to almost 8 m, with a standard deviation of 1.1 m.

In Trial 2, in which we allow the dike to extend toward the surface, the best- fitting model consists of a 7.6-km-high dike with 1 m of opening, equivalent to a reasonable overpressure of ~8 MPa. Allowing height to vary significantly improves the fit to the data, particularly at the western end of the leveling line where the model captures the upward inflection of the data. In this formulation, the marginal distributions demonstrate that dike opening is well determined, realistically ranging

103 from near-zero to around 5 m, whereas dike height is more poorly constrained and can vary over almost the entire parameter space.

If we allow the dike to extend laterally to the northeast of the seismic keel

(Trial 3), the best-fitting model is a dike 3.5 km in length with 2 m of opening, corresponding to an overpressure of ~20 MPa (Figure 30). Standard deviations of the marginal distributions (Table 7) suggest that the dike length is poorly determined in this case.

If dike opening, height and length are all allowed to vary (Trial 4), maintaining a bottom depth of 9 km, the optimal solution is a dike with only 0.25 m of opening, corresponding to an overpressure of 1 MPa. This solution fits the data only marginally better than Trial 2 (see weighted-r2 values in Table 7).

Allowing the dike’s vertical extent to vary over the entire upper crust (0-20 km) results in a 0.5-km-tall dike very near the surface that has an unrealistically large overpressure of 100 MPa (Trial 5), and attempts to exactly fit the W-shaped deformation at the western end of the leveling line. This trial also fits the data only marginally better than Trial 2.

4.7 Discussion

4.7.1 Best geometry

Constraining the dike strictly to the keel of seismicity results in a solution with a high overpressure (33 MPa), about three times the expected value. In addition, the relatively poor fit of this solution to the data, particularly at the western end of the leveling line, suggests that either some other process is occurring, or that the dike

104 extends beyond the seismic keel. Given the overwhelming evidence that a dike intrusion did occur at this location, we pursue the latter. In all trials, dike opening was between double and an order of magnitude larger than the 13-cm opening proposed by

Langbein et al. (1995). Allowing the dike length to vary in any of the described trials

(Trials 3, 4, 5) does not significantly improve the fit to the data, suggesting that the dike probably coincides with the seismic keel in terms of lateral extent. That is, the data are not sensitive to the length of the dike. Varying dike height (Trial 2), however, significantly improves the weighted-r2 value by up to 25% (Table 7), and results in an optimal model that has a reasonable amount of opening (1 m) and associated overpressure (~8 MPa). The lack of seismicity at depths shallower than the seismic keel is then perplexing. One possibility is that this portion of the crust is less seismogenic because of the presence of fluids or fractures, thus allowing dike emplacement without significant accompanying seismic activity.

Significantly, none of the trials with realistic overpressures (Trials 1-4) were able to mimic the W-shaped portion of the uplift profile at 10-13 km from CASA.

Given the significant leveling errors of around 4 mm (one standard deviation) for each measurement, and cumulative errors of up to almost 2 cm at 16 km from CASA, the short wavelength changes in uplift probably carry little informational content. We propose that the manifestation of the dike at the surface is reflected by the longer wavelength upward inflection at the western end of the profile, and the W-shape is not significant.

By calculating the volume of each model realization, we can generate probability distributions for volume. Tested models with reasonable overpressures

105 generate volumes around 0.015 km3 (Figure 31), with symmetrical probability distributions ranging from 0 km3 to 0.03 km3, consistent with a small, probably basaltic intrusion. This volume is only about twice as large as Langbein et al.’s proposed model, which had a volume of 0.0094 km3.

The orientation of the seismic “keel,” and hence the proposed orientation of the dike, approximately align with the strike of the Inyo and Mono craters farther to the north (Figure 27), suggesting that this intrusive event represents a continuation of

NNE-SSW striking volcanism west of the Long Valley caldera.

4.8 Conclusions

We reinvestigate the possibility of a dike intrusion at Mammoth Mountain,

California in 1989 by inverting the State Route 203 leveling line data (1988-1992) for dislocation parameters and incorporating information from relocated earthquakes

(Prejean et al., 2003) and a Long Valley inflation model (Langbein et al. 2003). We find:

1) Constraining the dike to exactly coincide with a plane of seismicity beneath

Mammoth Mountain located at 7-9 km depth and 2.3 km in horizontal extent,

yields a dike with 3.3 m of opening, corresponding to 33 MPa overpressure –

considered high for a dike.

2) Allowing the dike to extend toward the surface significantly improves the fit to

the data and results in a dike that extends 7-km vertically with ~ 1m of opening

and an overpressure of ~8 MPa, more typical of basaltic intrusions (Rubin,

1995). If the intrusion did intrude closer to the surface than the 7 -9 km-deep

plane of seismicity, intrusion shallower than 7 km depth was aseismic.

106

3) The short wavelength displacements at the west end of the leveling line have

magnitudes similar to measurement error, suggesting minimal informational

content. The dike intrusion is reflected in the leveling data by the upward-

facing inflection at the western end of the line.

4) Volume estimates for all realistic models were in the range of 0.009 km3 to

0.017 km3, consistent with intrusion of a small magma body.

107

4.9 Tables

Table 7: INVERSION RESULTS

Trial Dike Best fit MCMC MCMC σ Dike Over- Weighted r2 parameters median Volume pressure 0 N/A – Prolate - - - - - 39.2336 spheroids only 1 opening 3.3 m 3.4 m 1.1 m 0.015 km3 33 MPa 34.1160 2 opening 0.98 m 1.1 m 0.69 m 0.017 km3 8.5 MPa 25.6819 height 7.6 km 7.0 km 1.7 km (depth=9km) 3 opening 2.0 m 2.0 m 1.8 m 0.014 km3 20 MPa 34.0326 length 3.5 km 3.1 km 3.5 km 4 opening 0.25 m 0.97 m 1.1 m 0.009 km3 1.0 MPa 23.8709 height 7.1 km 6.7 km 1.8 km length 4.9 km 2.6 km 1.7 km (depth=9km) 3 5 opening 3.0 m 1.5 m 1.4 m 0.007 km 100 MPa 22.7981 height 0.60 km 2.9 km 3.7 km depth 3.7 km 5.8 km 3.6 km length 3.9 km 2.8 km 2.1 km Notes: Results of inversions, consisting of a simplex search algorithm (best fit) followed by Markov Chain Monte Carlo (MCMC) simulation (estimation of σ). Unless specified, height = 2 km, depth = 9 km, length = 2 km, dip = 90o, strike = 17.3o. Center of models 1 and 2 is at -119.037o W, 37.615o N. Prolate spheroids are from Langbein et al. (2003): Spheroid 1 is located at (-118.9239, 37.6817), volume = 0.0288 km3, azimuth = 0o, plunge = 80o, depth = 6 km ratio of min to max stress = 0.75; Spheroid 2 is located at (-118.9089, 37.6417), volume = 0.0184 km3, azimuth =250o, plunge = 50o, depth = 12 km, ratio of min to max stress = 0. Table 7: Mammoth Mountain – Inversion results

108

4.10 Figures

Figure 27: Location of study with respect to Long Valley caldera, CA. Red line represents caldera rim, small orange lines represent craters, the green asterisk is the location of the CASA benchmark, just west of the Casa Diablo geothermal plant, and green dots represent prolate spheroid inflation sources from Langbein et al. 2003 at 6 and 12 km depth, detailed in the note of Table 1. Yellow dots represent points where CO2 has been detected and the blue line marks the Route 203 leveling line. Base map from Battaglia et al. (2003). NAD 1927 projection is used throughout.

109

Figure 28: Close up of Mammoth Mountain area and seismicity from May 1989 through January 1990. Black dots show relocations of earthquakes in the 1989 seismic swarm, shallower than 7 km and pink dots show the deeper seismicity between 7 and 9 km depth, i.e. the seismic keel (Prejean et al., 2003). The orange triangle marks the summit of Mammoth Mountain, the red line represents the caldera rim, and the blue dots show the benchmark locations along the Route 203 leveling line. The trace of the dike at the surface is shown in green, assuming that the dike is coincident with the seismic keel. Base map from Battaglia et al. (2003).

110

Figure 29: Observed data (blue) and deformation model (pink) of Long Valley caldera based on inflation of two prolate spheroids at 6 km and 12 km depth (see Figure 27) from Langbein et al. (2003). Green dots represent benchmarks affected by motion at the Casa geothermal plant and are not used in the inversion. Figure after Langbein et al., (1995).

111

Figure 30: Best fitting model and marginal distributions of the inversions for all trials (Table 7). In the best fitting models, blue line represents data with 1-sigma error bars and red dashed line represents the best fitting model. Horizontal axis in marginal distributions is scaled to show all allowed values of the parameter. Red dashed line on marginal distribution shows value of best fit, and the cyan solid line shows the median value. Allowing the dike height to vary (Trial 2) significantly increases the goodness of fit. In both cases, the model over-predicts uplift at the eastern end of the leveling profile.

112

Figure 31: Volume distributions of inversions for Trials 1 through 4. For each trial, the red dashed line is the volume of the best-fitting model and the cyan line is the median volume.

113

5 Investigation of Deformation Sources for the 2000 Miyakejima, Japan, Intrusion and Seismic Swarm

Data for this chapter came from a variety of sources. GPS data was provided by

Shinichi Miyazaki at the Earthquake Research Institute of Japan and Sue Owen at the

Jet Propulsion Laboratory, California. In addition, Yosuke Aoki of the Earthquake

Research Institute, Japan and David Shelley from the USGS, Menlo Park helped obtain seismic data. This project would not have been possible without their help.

Paul Segall advised this project.

5.1 Abstract

On June 26, 2000, a large seismic swarm began beneath the west coast of

Miyakejima, Japan, an active volcanic island in the Izu arc. After a week-long propagation of seismicity to the northwest, seismic activity continued for approximately two months between Kozushima and Miyakejima islands. The swarm was accompanied by eruptions and caldera collapse on Miyakejima, as well as continued crustal deformation measured by the GEONET GPS network. Based on observed deformation and locations of hypocenters, it has generally been assumed that dike propagation from a magma chamber beneath Miyakejima was responsible for the swarm. Although several models of varying complexity have been proposed to explain the deformation, there is little consensus on a geologically motivated model that explains both the long-lived deformation and seismicity.

114

Any model proposed must address three questions based on observations of the data. (1) Why are the cumulative GPS derived displacements after the first week so small (< 5cm) relative to ~80 cm of final displacement if the dike propagated to its full lateral extent during this time? (2) Why did the dike stop propagating laterally after seven days if related deformation continued for two more months? (3) And finally, was the dike located above or within the main cloud of seismicity? In order to determine viable sources of observed deformation and seismicity, we tested several geologically motivated models based on observed displacements, hypocenter locations, overall tectonic setting, and pre-event seismic data. Our results show: (1)

Aseismic slip on a fault aligned with pre-event seismicity between Kozushima and

Niijima explains the anomalous motion at the Shikinejima station; (2) The dike may have stopped propagating after the first week when it intersected this strike-slip fault zone; (3) Models that tie the dike directly to Japanese Meteorological Agency (JMA) hypocenter locations do not fit deformation data as well as those models that allow the dike to extend shallower than the main cloud of seismicity; (4) The preferred model for the full event includes a dike that extends from ~2 km to 13 km depth, coincident with the extent of hypocenters located by Earthquake Research Institute of Japan

(ERI) (Sakai et al., 2001).

5.2 Introduction

The Izu arc of Japan (Figure 32) is the result of subduction of the Pacific Plate beneath the Philippine Sea Plate and consists of a series of volcanic islands trending

NNW-SSE. The arc itself is moving northwestward at a rate of 4-5 cm/yr as the

Philippine Sea Plate subducts beneath southern Japan and collides with the Japanese

115 mainland at the Izu-peninsula. Miyakejima (Miyake Island) is a small basaltic island that has developed over the past ten thousand years. Similar to other basaltic shield volcanoes, Miyakejima has regularly experienced dike intrusion and minor eruptions with a recurrence interval of approximately twenty years during recent history (Saito et al., 2005).

On June 26, 2000, one the largest seismic swarms in Japanese history initiated beneath the west coast of Miyakejima, and began propagating northwestward toward

Kozushima (Kozu Island). On the morning of June 27, a small submarine eruption was recorded approximately 1.5 km off the coast of Miyakejima (Irwan, et al., 2003 and

2006) while seismicity continued to migrate northwestward and downward (Figure

33). By July 2nd, the earthquake swarm had nearly reached Kozushima, but displacements measured by continuous GPS on Kozushima and Niijima were less than

5 cm, suggesting an initially narrow or shallow dike (Figure 34). The seismic swarm persisted for nearly two months, consisting of more than 7,000 M ≥ 3 shocks, until the rate of seismicity began to rapidly decay after a small eruption on Miyakejima on

August 18 (Yamaoka, et al., 2005). During this time period there were a total of five earthquakes with magnitudes greater than 6.0, and five phreatic eruptions on

Miyakejima (Figure 34). Hypocenters are concentrated in a vertical plane that extends from Miyakejima to just southeast of Kozushima (Figure 33). A caldera progressively formed on Miyakejima’s summit beginning after an eruption on July 9th and continuing throughout the event, eventually reaching 1.6 km in diameter and a depth of 400 m (Fujita et al., 2002). Modeling of the deformation on Miyakejima recorded by tiltmeters and GPS suggests the presence of a deflating magma chamber at around

116

4-6 km depth (Ueda et al., 2005). Continuous GPS stations on local islands eventually recorded dramatic horizontal crustal deformation (Figure 34) including an increase of the distance between Niijima and Kozushima of more than 80 cm (e.g., Toda et al.,

2002). Erupted products were basaltic and amounted to only 0.02 km3 (Saito et al.,

2005), a significantly smaller volume than the caldera itself.

This event is of particular interest because of its longevity (over two months) and the observation that the dike appears to have stopped propagating laterally after seven days but deformation and seismicity continued for another two months. Study of the Miyakejima, 2000 activity may thus deepen or change our understanding of how mafic magma is emplaced in the crust. Because lateral dike propagation induces a change in stress as it moves, it is not surprising that the intrusions are marked by an associated propagation of earthquake clusters (e.g., Rubin et al., 1998). For example,

Dieterich’s (1994) work on earthquake clustering quantifies how changes in stressing rate lead to changes in the local rate of seismicity. Assuming in this case that the initial migration of seismicity away from Miyakejima (Figure 33) marks the approximate location of the dike tip, the intrusion appears to reach its full extent during the first week of activity (June 26 – July 2). If this is the case, a mechanism needs to be derived by which significant crustal deformation could continue after the dike had stopped propagating laterally. This study will examine the event at two different temporal scales (first week and full two month periods) in order to determine if and how the source of the deformation may have changed over time.

Any model for this event needs to address three major questions that arise from initial examination of the seismic swarm and GPS data:

117

1) Why is there so little deformation during the first week of the event, when

the dike appears to have reached nearly its full horizontal length? Previous

studies have modeled this minor deformation with a dike that does not

match the initial extent of seismicity (e.g., Ozawa et al., 2004; Murase et

al., 2006).

2) Why did the dike appear to stop propagating laterally, given that GPS

stations recorded dramatic continuously increasing deformation for two

more months?

3) Where is the magma in relation to the seismicity? Is the dike collocated

with the main cloud of seismicity or is it located in the aseismic zone above

the JMA hypocenters?

5.3 Previous Work

Several models have been proposed to account for the crustal deformation around Miyakejima during the 2000 activity. All assume some combination of dike intrusion and magma source deflation beneath Miyakejima caused the observed displacements, and many also incorporate seismic and aseismic slip on presumed faults. For example, Toda et al. (2002) proposed that the final “dog-leg” shape of seismicity resulted from the transfer of shear stress from an intruding dike to surrounding vertical faults. In addition to a magma chamber beneath Miyakejima, many models suggest an additional deep magmatic source between Miyakejima and

Kozushima was “awakened” by overlying activity, though proposed degrees of involvement of this source vary widely (Furuya, et al., 2003; Ito and Yoshioka, 2002;

Murase, et al., 2006; Nishimura, et al., 2001; Yamaoka, et al., 2005). The proposal of

118 a secondary source is primarily motivated by the discrepancy between the estimated volumes of the dike (~ 1 km3) and the deflation of the Miyakejima magma chamber (~

0.1 to 0.2 km3) in the various studies. In addition, study of GPS data during a time interval prior to the Miyakejima 2000 event shows a slight extension between Kozu and Niijima Islands, which was interpreted to be a possible magmatic source (Ozawa, et al., 2004). However, the process by which such a source could be activated by overlying activity is not clear.

There are two mechanisms that could account for continued crustal deformation following cessation of the swarm elongation. First, Toda et al. (2002) and Yamaoka et al. (2005) propose that the dike propagated during a short time period and then continued to open for the duration of activity, resulting in 20 m and 10 m- thick intrusions respectively. This would require an increase in magma-pressure, while the increase in dike volume presumably led to a decrease in pressure. For this to occur there would need to be some impediment that limited dike growth forcing the dike to continually open, despite the increasing stress concentration at its edges. In order to explain their model, Yamaoka et al. (2005) assert that a second magmatic source beneath the dike also contributed to its continuous growth. Other studies assert that prolonged crustal deformation is due to the intrusion and expansion of multiple dikes during different stages of the activity (Furuya, et al., 2003; Murase, et al., 2006;

Nishimura, et al., 2001). Murase et al. (2006) for example proposed that a series of three dikes opened at different times between Miyakejima and Kozushima, fed first laterally by a magma chamber beneath Miyakejima and later vertically by a deeper magma supply located between the two islands. While it is theoretically possible that

119 the dike intersected a pressurized underlying magma chamber between Kozushima and Miyakejima, we do not see a sudden break in slope early in the GPS station timeseries (Figure 34) that would indicate an increased magmatic flux. Another possibility is that the dike extended vertically, which would lead to increased opening, at constant or potentially declining magma pressure.

Much research also specifically addresses magma movement during the first two days of activity (June 26th and 27th). Analysis of seismicity and GPS data for the first two days of the event, prior to a small submarine eruption, resulted in a model consisting of three successive magma emplacements trending northwest away from

Miyakejima (Irwan et al., 2003; and 2006). These intrusions are modeled as being within 1 km of the surface and relatively small (1-5 km) as compared to estimates of the total intrusion (20-30 km). Ueda et al. (2005) likewise utilized tiltmeter data from

Miyakejima as well as GPS data to develop a three-period model during which three dikes were intruded from two separate sources.

Missing from this previous work is a detailed examination and correlation of seismicity and displacement during both the first week of activity, when magma appears to have propagated rapidly northwestward, and throughout the remainder of the event. Pre-event seismicity has also been largely over-looked and may provide information about pre-existing local structures. Additionally, in order to determine whether and how the source of deformation changed over the nearly two-month period, it is necessary to compare deformational patterns over a large geographical area at short (one week) and long (~ two month) time-scales. The goal of this study is to use observations from the seismicity and inversion of GPS data to determine

120 realistic, geologically motivated mechanisms that can account for the long-lived surface deformation near Miyakejima.

5.4 Data and initial observations

5.4.1 Continuous GPS and Deformation

By 2000, the Geographic Survey Institute (GSI) of Japan had installed more than 240 telemetered continuous GPS known as the GEONET network (Figure 35A).

Daily solutions for all stations were calculated by GSI and by the Jet Propulsion Lab in California with the processing program GIPSY (Owens, S., personal communication). Solutions used in this study were provided by JPL. Position uncertainty is reported as 3-5 mm in the horizontal and 7-9 mm in the vertical.

Given that the 2000 intrusion occurred under water between the islands, the spatial coverage in this area is not ideal, but displacements related to the intrusion were recorded on surrounding islands and at stations on mainland Japan, more than

100 km from the source. Because the decay of displacements with distance from the source gives information about source depth, these far field displacements are used in the inversion of GPS data to solve for the dike geometry. Stations with obviously anomalous behavior (such as random excursions of several centimeters) were removed. In all, daily solutions for 61 stations were used in the inversion of the full time period. All displacements were calculated with respect to a fixed Station 0241, located on the far north side of Honshu (Figure 32).

In order to distinguish deformation specifically related to the intrusive event, three steps were taken to “clean” the daily solution timeseries. First, reference frame

121

“jitter” was removed with a Kalman filter (Murray, 2003), in which a reference frame correction was calculated and removed from all stations for each day, thereby decreasing the amount of day-to-day noise in the data. Second, stations used in the inversion were located on four different tectonic plates, necessitating the removal of secular motion for each individual station. An averaged daily rate of motion was calculated at each station from a three-year period spanning June 1997 through mid-

June 2000. Secular displacements over the ~ 2 month period of the seismic swarm ranged from 0.4 to 7.6 mm depending on station location (Figure 36) and were subtracted from the displacements. Finally, five ~ M 6 earthquakes (Figure 33) occurred over the course of the event, which created significant steps in the station timeseries. In order to isolate displacements related to dike propagation and possible aseismic slip, these earthquakes were removed from the timeseries. This was done by first subtracting the displacement at each station on the day of earthquake and then adding back an averaged daily displacement based on the four days following the earthquake, thus creating a smoothed timeseries. This method assumes that the displacements are approximately linear at the timescale of a few days. It should be noted that the Tokai slow slip event initiated during late June of 2000, but based on the work by Miyazaki et al., (2006) which deconvolved displacements for the two deformation sources, significant deformation related to slow slip was not recorded by the stations used in this inversion.

122

5.4.2 Seismicity

A dense network of seismometers and some ocean-bottom seismometers covers most of Japan (Figure 35B). Initial hypocenter locations and phase data for the

2000 earthquake swarm were taken from the Japanese Meteorological Agency (JMA)

Catalogue. In addition, 54 ocean-bottom seismometers (OBS’s) were deployed by the

Earthquake Research Institute of Japan (ERI) (Sakai et al., 2001) on July 1, 2000, after the intrusion and seismic were underway (Figure 35C). Hypocenter depths vary significantly between these two datasets. Presumably the ERI locations are better constrained because the OBS stations were incorporated. The line of seismicity at 13 km depth in the ERI data and the presence of earthquakes above the datum

(presumably sea level), however, leave questions about their velocity model (which has five layers). In order to glean as much information as possible, we examine each seismic dataset in turn.

5.4.3 JMA Catalogue

Assuming that earthquakes are a reflection of stress changes induced by dike propagation and inflation and/or aseismic slip, it is important that hypocenters are located as accurately as possible so that source models can be better constrained.

Given that the earthquakes depths in the JMA catalogue` are probably poorly constrained, the relocation of these events primarily gives information about the horizontal locations of earthquakes. Relocation was accomplished by using the

HypoDD earthquake relocation program (Waldheuser and Ellsworth, 2000).

123

5.4.3.1 Relocating Earthquakes

The HypoDD program relocates hypocenters by using a “double-difference” algorithm, where the first difference refers to the difference between the observed and predicted travel times. The second difference refers to the difference in the travel times of seismic waves for two different events to the same station. If the events are closely spaced, the residual in the travel time can be reasonably attributed to the distance between the two events, rather than to variability in crustal structure.

All earthquakes that had preliminary hypocenters were used, and P and S phase data were chosen from the observations, although negative travel times were removed, as they are impossible. Each station observation (P or S) has an assigned a priori weight in the JMA catalogue. Only those station observations with values of 1.0 (on a scale from 0 -1) were used in the hypoDD relocation, resulting in a total of ~12,500 hypocenters used in the relocation (Figure 37A).

A one-dimensional velocity model for crust underlying the Izu islands was obtained from the results of a reflection-refraction experiment described in Takahashi et al. (1998). The crust in this region is approximately 23 km thick and can be divided into eight layers based on P-wave velocity. The ratio between P-wave and S-wave velocities is fixed at 1.73.

Given the large size of this swarm, we used the least-squares residual (LSQR) approach rather than the more computationally intensive singular value decomposition

(SVD) method. Unfortunately, the LSQR method results in unreliable error estimates for the relocated hypocenters. In order to get an estimate of location errors, we completed a bootstrap analysis of the relocations (Waldhauser and Ellsworth, 2000),

124 examining only the first 1000 events in order to make the problem computationally feasible. 100 Bootstrap samples were created from the travel-time residuals for the relocated hypocenters, as output from hypoDD. Output from this investigation is shown in Figure 38. We define the uncertainty of the hypocenter relocations as the standard deviation of the distribution of location differences from the bootstrap analysis. Errors are thus 0.53 km in the east-west direction, 0.450 km in the north- south direction, and 0.64 km in the vertical, although as stated before, vertical errors are probably significantly larger because of the sparse station distribution of JMA seismometers in this region. In addition, the first 1000 events took place during the first two days of activity and are thus clustered close to Miyakejima. These earthquakes may thus be more accurately located in the JMA catalogue than later earthquakes that fall between islands, i.e. those located at 10-20 km depth.

5.4.3.2 Observations from Relocated JMA Seismicity

Several interesting features can be observed in the relocated seismicity (Figure

37B): (1) There is a prominent plane of seismicity that is presumably associated with dike intrusion, striking ~130o; (2) This central plane of seismicity is vertical or nearly vertical; (3) In cross-section there is a distinct U-shape to the hypocenters, in which earthquakes closest to Miyakejima are shallow, in the top 5 km of the crust, and earthquakes in between Miyakejima and Kozushima are located near the base of the crust, at about 10 to 20 km. (3) There is a moderately dipping “tail” of earthquakes that extends from the central zone of seismicity to just below Kozushima and Niijima, suggesting that there may be a secondary structure (such as a pre-existing fault) in this

125 region. (4) There is a very shallow horizontal cloud of earthquakes off the west coast of Miyakejima, most likely marking the point where the dike intersected the surface, causing the small submarine eruption on June 27th; (5) Directly beneath Miyakejima, there is a small cluster of earthquakes at around 4-5-km-depth. Examination of the distribution of hypocenters over time (Figure 33) shows that seismicity initially propagated northwestward and down during the first week, but almost all subsequent activity was confined to a region between Kozushima and Niijima at 12 - 20 km depth.

5.4.3.3 Focal Mechanisms

The National Research Institute for Earth Science and Disaster Prevention

(NIED) of Japan calculates and provides focal mechanisms for JMA catalogue entries with magnitudes larger than 3 (Figure 39A). There are a preponderance of normal and strike-slip mechanisms. Figure 39B shows the `(direction of greatest tension) and P- axes (direction of greatest compression) for earthquakes with M ≥ 4. T-axes of earthquakes between Miyakejima and Kozushima are oriented perpendicular to the presumed strike of the plane of seismicity and approximately horizontal (Figure 39C), consistent with dike emplacement. T- and P-axes for earthquakes located between

Niijima and Kozushima have a different average orientation than those located near the main plane of seismicity, indicating that the stress state near these islands may be different from that closer to Miyakejima.

5.4.3.4 Pre-event Seismicity

In addition to earthquakes from the 2000 seismic swarm. We examined JMA catalogue hypocenter locations for earthquakes in this region during a ten-year period

126 before the 2000 event (Figure 40). These hypocenters fall on a plane with a strike of

~36o, perpendicular to the strike of the 2000 seismic swarm. Many of the focal mechanisms in this zone of seismicity have one nodal plane parallel to the zone of seismicity showing left lateral motion, consistent with the focal mechanisms of earthquakes that occurred in this area during the 2000 swarm. There is also considerable evidence for normal faulting just north of Kozushima.

5.4.3.5 ERI Dataset

Sakai et al. (2001) of ERI examined earthquakes recorded at 3-component stations on Kozushima, Niijima, Shikinejima, and Toshima in addition to data recorded by ocean bottom seismometers (OBSs) deployed on July 1, 2000. We will refer to the resultant hypocenter locations as the ERI dataset. Sakai et al. relocated this seismic dataset using a five-layer velocity model (Figure 41). Errors for the relocated earthquakes are estimated as 0.54 km in the vertical and 0.17 km in the horizontal. The horizontal locations of earthquakes in the ERI dataset approximately match those from the relocated JMA catalogue (Figure 41A), but the earthquake depths are significantly different (Figure 41B and C). During the first week of activity, hypocenter locations follow a descending pattern similar to the JMA catalogue earthquakes, but the swarm does not descend as deep (only to ~ 13 km).

Given that prior to July 1 the ERI dataset is simply a subset of the JMA dataset (OBSs had not yet been deployed), this discrepancy in hypocenter depth is intriguing. It is conceivable that the difference is a result of which stations are being used or are potentially the product of different velocity models. After the first week, hypocenters

127 were concentrated in a zone between Kozushima and Niijima extending from 2 to 13 km depth. The base of this swarm is suspiciously flat and likely an artifact of a layer boundary in the employed velocity model, which has layer boundaries at 2.5 km and

12.5 km (Sakai et al., 2001). Concentrated seismicity beneath Miyakejima extends from the surface down to ~3-5 km. In addition, Sakai et al. (2001) report that hypocenters sometimes migrate from deeper to shallower depths, possibly indicating that magma is being supplied from a deep source between Miyakejima and

Kozushima, rather than from Miyakejima itself.

5.4.3.6 Summary of Observations from Seismicity

In summary, the seismicity associated with this event tells us a great deal about the potential geometries of deformation sources. There is a vertical planar alignment of hypocenters that initiated at shallow depths beneath Miyakejima, propagated downwards during the first week of activity and finally remained between Miyakejima and Kozushima through August. Depending on which hypocenter dataset is used, the bulk of planar seismicity occurred at either at 10 - 20 km depth or 2 – 13 km. Because the ERI locations employed an extensive OBS array the shallower locations are likely to be more accurate. In addition, based on both the pre-event and syn-event seismicity, it is very likely that there is a pre-existing, left-lateral fault zone extending through Kozushima and Niijima, oriented perpendicular to the dike.

5.5 Modeling GPS Deformation

As noted in several previous studies, a simple model including a vertical dike

(aligned with the strike of the main body of hypocenters) and a deflating isotropic

128 point (“Mogi”) source beneath Miyakejima does not adequately fit the observed deformation. The best fitting model consisting of a uniformly opening dike, which matches the horizontal location of the seismicity, and a Mogi source alone is shown in

Figure 42. At the station on Shikinejima in particular, even the sense of motion predicted by such a model is grossly incorrect. Given that there is no reason to assume that this station is unstable and the apparent presence of a pre-existing seismic zone to the northeast, it is likely that there is some other process occurring, such as aseismic slip in the Kozushima-Niijima fault zone. In addition, displacements are generally under-predicted, particularly at Kozushima.

The modeling process was divided into three steps. We first inverted the GPS displacements for the full event (June 26 – August 31), testing geometries based on geological and physical constraints as well as the hypocenters of the relocated earthquakes. We assume an elastic half-space with dike and fault sources modeled as uniform rectangular dislocations and the magma chamber modeled as a Mogi source.

Because it is not clear how the earthquakes are temporally and spatially related to changes in model geometry, hypocenter locations serve to constrain model parameters but are not explicitly included in the inversion. Second, we fit the displacements observed during the first week of activity (June 26 – July 2), when the seismic swarm propagated from Miyakejima to Kozushima. For this phase we hold most parameters fixed based on the best-fitting model for the full event, allowing only the length and vertical extent of the dike, the amount of dike opening, and fault slip to vary. This restrictive methodology is used because displacements during the first week are small

(< 5 cm) and may be significantly affected by random errors as well as displacements

129 due to the earthquakes themselves, making an inversion for more parameters poorly constrained. Lastly, we use a forward model for dike growth to compare the observed

GPS timeseries for the entire event to those predicted by candidate model geometries.

5.5.1 Constraining the Model

All models consist of a dike extending from Miyakejima, a deflationary source beneath Miyakejima, and a fault at the northwest end of the seismic swarm.

5.5.1.1 Dike

The model for a uniformly opening rectangular dislocation has eight parameters: length, height, depth to the bottom edge, strike, dip, center location (x,y), and amount of opening. For the full time period we assume that the orientation and horizontal location of the dike are given by the lateral extent of the seismicity. Strike is fixed at ~130o and the eastern end of the dike is fixed to coincide with the eastern extent of the seismic swarm, just off the southwest coast of Miyakejima. We likewise fix the dip at 90o because both the JMA and the ERI relocated hypocenters show a vertical plane of seismicity. These assumptions leave only four free parameters: length, height, depth to bottom edge, and opening. The ratio of dike opening to height

(assuming height is less than length) was limited by a pressure constraint given by the equation:

(8) where w = opening, l = minimum dike dimension, P is the pressure within the dike, S is the dike-normal stress, µ is the shear modulus and ν is Poisson’s ratio. Given

130 standard values of the shear modulus and Poisson’s ratio (30 GPa and 0.25 respectively), typical values of overpressure (P-S) based on field observations are in the range of 1 – 10 MPa for basaltic dikes (Rubin, 1995). We conservatively constrain dike overpressure to be between 0.1 and 100 MPa, noting that overpressures greater than 20 -30 MPa should not be considered realistic . The models proposed by Toda et al. (2002) and Yamaoka et al. (2005), which include dike openings of 20 and 10 m and minimum dimensions of 5 km and 10 km, would thus be associated with dike overpressures of 80 and 40 MPa respectively, making such large dike openings implausible.

5.5.1.2 Strike-slip Fault

A number of lines of evidence indicate that a vertical, left-lateral fault zone lies along the strike of Kozushima and Niijima:

1) Pre-event seismicity is concentrated in a nearly vertically plane aligned with

Niijima and Kozushima (Figure 40).

2) NIED focal mechanisms for large pre-event and syn-event earthquakes between

Niijima and Kozushima indicate a near-vertical, left-lateral, strike-slip fault

oriented at 36o (Figure 39 and Figure 40).

3) Modeling the effect of dike opening on the surrounding stress regime (using

Coulomb 2.0) suggests that left-lateral slip is feasible perpendicular to the end of

the dike on the northeast side (Figure 43).

4) The fact that the dike stopped propagating laterally may be related to its

intersection with this pre-existing fault zone.

131

We model the fault with a uniformly slipping rectangular dislocation where opening and normal slip are constrained to be zero. We restrict the fault strike to coincide with the vertical plane of pre-event seismicity and allow only left-lateral strike slip motion. Because the zone of pre-event seismicity was fairly broad and the stress modeling indicates that the most likely area for faulting is at the dike tip, we require the southwest endpoint of the fault to coincide with the northwest endpoint of the dike. The vertical extent of the fault was allowed to vary over the entire crust, from the surface to 20 km. Fault slip was constrained by reasonable values of stress drop between 0 to 100 MPa, where stress drop on a rectangular fault (ΔSf) is empirically defined as:

(9)

(Chinnery, 1964) where l is the minimum fault dimension.

Based on the hypocenter locations (both JMA and ERI), we have developed four models for the full event (June 26 – August 31), ranging from highly constrained to minimally constrained. Dike length is allowed to vary in all four models, given that the western of extent of seismicity does not have a clear end. In Model F-A, the dike is constrained to have a height of 10 km and a depth of 20 km, such that the dike lies within the central plane of seismicity observed in the relocated JMA hypocenters. In

Model F-B, the dike is constrained to have a base that coincides with the base of the

ERI dataset, at 13 km (Saito et al., 2001). In Model F-C, neither the height nor the depth is fixed and the vertical extent of the dike is allowed to vary over 20 km, nearly the entire crustal thickness. In Models F-A, F-B, and F-C, the fault plane is

132 constrained to be perpendicular to the dike, coincident with the pre-event seismicity.

F-D is identical to F-C but tests the hypothesis that aseismic slip occurs on a fault that is oblique to the dike, parallel to the ~N-S striking line of earthquakes to the northwest of Niijima.

5.5.1.3 Mogi source

A Mogi source has four parameters: location (x, y), depth, and volume change.

Because of the caldera collapse, the location of the deflating magma chamber can be pinpointed. In addition, the relocated hypocenters reveal a cluster of earthquakes beneath Miyakejima at about 4 km depth and several previous studies place the Mogi source at 4 – 4.6 km depth (e.g. Nishimura et al., 2001; Ozawa et al., 2004; Yamaoka et al., 2005) based on GPS and tilt data. Negative volume change beneath Miyakejima has been constrained by a combination of mass flux analysis and inversion of displacements resulting in a presumed volume change of ~ 0.13 km3 (Yamaoka et al.,

2005). The large displacements (up to ~ 80 cm) recorded by stations on Miyakejima are the primary record of deflation on the island, but these stations were also probably affected by volcanic eruptions, caldera collapse and shallow magma propagation early in the event (e.g. Irwan et al., 2003 and 2006). We have thus chosen to significantly down weight the Miyakejima stations by multiplying their associated variances by a factor of 10 and fixing the depth of the Mogi source based on the work of Yamaoka et al., (2005) (4.6 km), inverting only for the change in magma chamber volume.

133

5.5.2 Inversion Methods

The inversion process was split into two phases: a weighted simplex search algorithm to find an optimal solution, and a Markov Chain Monte Carlo simulation that refines the solution and generates a posterior probability distribution of the model that can be used to calculate a model error.

5.5.2.1 Weighted Simplex Search Algorithm

The algorithm uses the Matlab program fminsearch, a Nelder-Mead simplex search algorithm that seeks to minimize a given function (Lagarias et al. 1998). In this case, the algorithm is set to maximize the L2 formulation for the likelihood of the model parameters (s) given the data (d):

(10) L(s | d) (2 )−n / 2 (| |)−1/ 2 exp (rT −1r)/2 = π Σd [− Σd ] where n is the number of data, r is the difference between observed and calculated

€ displacements at each station, and Σd is the data covariance matrix. For simplicity, we minimize the negative logarithm of L, ignoring the constants:

T −1 (11) −ln(L) = −(r Σd r)/2

Far field stations perpendicular to the strike of the dike (Figure 44) were given more weight in the inversion€ (by a factor of ten) because they should provide information about dike depth. This also has the effect of down-weighting the (small) displacements on the Izu Peninsula that may be influence by the Tokai slow slip event

(Miyazaki et al., 2006). Because the algorithm can fall into local minima, it was run

150 times for each model beginning at randomly generated points in the parameter space.

134

5.5.2.2 Metropolis-Hastings Markov Chain Monte Carlo Simulation (MCMC)

The Metropolis-Hastings MCMC algorithm is frequently employed as a method for determining the spread of feasible models. The basic algorithm is as follows:

1) Begin at a random initial guess in the parameter space, so, with likelihood Lo

given the data.

2) Take a step to a new trial location in the parameter space, strial. and calculate its

likelihood given the data, Ltrial.

Ltrial 3) Let " = Lo

4) If α > k, where k is a randomly chosen number from U(0,1), then strial is ! accepted as then next step in the chain, s1. If α ≤ k, strial is rejected and s1 = so.

Steps 2 through 4 are then repeated until the algorithm builds up a smooth posterior probability distribution, viewed as individual marginal distributions for each parameter, where a greater density of realizations at a certain value indicates the most likely value of a given model parameter.

In this study, the likelihood function is a variation of (10):

$n $1 (12) 2 2 2 2 2 T $1 L(s | d," ) = (2#" ) (| %d |) exp[($1/2#" )(r %d r)] where we assume the data errors have a Gaussian distribution with mean 0 and

! 2 2 covariance matrix σ Σd. It is necessary to estimate σ for each model because given that we are using simple source geometries the model error may be larger than the data error. The MCMC algorithm is then run twice for each model – once to determine the value of σ2, and then again with σ2 fixed to generate the marginal distributions for the

135 model parameters. This two step process prevents bias in the estimate of σ2

(Kitanidis, 1996). We use the best model estimated from the simplex search algorithm as the initial point for both MCMC runs. For each model we compute 100,000 realizations per parameter, setting the step size such that ~25% of new steps are accepted.

5.6 Results

Results of the inversion process for Models F-A through F-D are summarized in Table 8 and Figure 45 through Figure 48. We use the weighted-r2 (X2) as a measure of the goodness of fit of each model defined as:

2 T −1 (13) Χ = (r Σd r)

Note that this value is not divided by σ2, as in the likelihood given in (12), so that we € may quantitatively compare results for different models that may have dramatically different values of σ2.

5.6.1 Full Event

The highly constrained model (Model F-A) fits the observed displacements poorly (Figure 45), particularly at stations on Kozushima and Niijima. This model tends to underestimate displacements in the near field, and overestimate displacements in the far field, as seen in the dike-normal profiles (Figure 45). Allowing the dike height to vary but holding the base at 13 km (Model F-B) results in a model that fits the data well and matches the dike-normal profile (Figure 46), i.e. the model does not systematically over- or underestimate displacements at near-field or far-field stations.

The top of the dike in this model extends to around 2-km depth, which matches the

136 vertical extent of seismicity in the ERI dataset. In addition, the dike in this model extends almost exactly to the edge of the pre-event seismic zone. Allowing the dike depth to vary as well (Model F-C, Figure 47), results in a weighted r2 value that is only marginally better than that of Model F-B, suggesting that Model F-B is optimal in that it both fits the data well and matches the ERI hypocenter distribution. Model F-D fits the displacement data approximately as well as Models F-B and F-C but the large dike overpressure and fault stress drop required to fit the displacements indicate that this is an unlikely solution. Model F-B is then the preferred model for the full event. The possible kinematics of this model are shown in Figure 49, where dike opening near the northwest tip is accommodated by left-lateral aseismic slip on a pre-existing fault.

The results of the MCMC portion of the inversion for each model are shown in the lower portions of Figure 45 through Figure 48 and Table 8. Values of σ2 vary from ~70 to nearly 500, indicating that (1) the models are over-simplified and/or (2) the error estimates for the GPS data, which are on the order of 3 mm in the horizontal and 6 mm in the vertical, are probably overly optimistic. Examining the marginal distributions for each parameter, one can see that some parameters such as dike opening are well constrained where as others including the fault parameters (fault slip in particular) tend to have more variability.

5.6.2 First Week

During the first week of the event (June 25 – July2, 2000), when the seismic swarm propagated northwestward, there was little deformation recorded at GPS stations other than those on Miyakejima. The signal-to-noise ratio for these

137 displacements is thus comparatively low, making inversion much more difficult. In addition, much of the deformation observed during the first week may have been caused by earthquakes that have not been completely subtracted from the timeseries.

Because of these difficulties, we have taken the approach of using the optimal geometry from the full event (Model F-B) to constrain the strike-slip fault geometry for the first week, inverting only for fault slip, dike opening and, in Model W-A, dike height and depth. Attempts to fit the first week of data without slip in the fault zone fit the data very poorly (Model W-O, Figure 50). It is plausible that the poor fit is due to small earthquakes and not due to aseismic slip within the fault zone between

Kozushima and Niijima, but modeling displacements due to all these small earthquakes is not possible at this juncture. Because the displacements on

Miyakejima dwarf those on surrounding islands during the first week, we again constrain the location and depth of the deflation source to that of Yamaoka et al

(2005). We additionally constrain the volume change to -0.02 km3 based on initial inversions. Only near-field stations were used in the inversions of first-week displacements.

We test two end-member models. In Model W-A, we assume that the dike propagated to its full vertical extent during the first week and then simply continued to open over the next two months. In the second (Model W-B), we allow the dike height, depth, and opening to vary, i.e. we assume that the dike both opened and propagated vertically over time. In both models the dike opening and fault slip are constrained to be less than those of the dike for the full time period in Model F-B.

138

The results of these inversions are shown in Figure 51, Figure 52, and in Table

8. Allowing dike height and depth to vary, (Model W-B), results in a dike that is very short and close to the surface (1-2 km depth), but has ~1 m of opening, associated with an overpressure of ~ 15 MPa. In contrast, requiring the dike to extend from near the surface to 13 km depth (as in Model F-B), results in a dike with only ~0.2 m of opening and a small overpressure of 0.3 MPa (Model W-A). Interestingly, the weighted-r2 value for the less constrained model is only slightly smaller (i.e. better) than the fully constrained model. For both models, the optimal solution has a dike length of 30 km, equal to that of models for the full event. The marginal distributions from the MCMC output for both models indicate how poorly constrained these models are. For example, based on the marginal distributions for Model W-B it is plausible that a dike 2-km-high dike extended from the Miyakejima magma chamber with base at 5 km depth (Figure 53A, Model W-C). We can however rule at a 2-km-high dike propagating with base at 10 km depth (Figure 53B, Model W-D). In order to determine which model (W-A or W-B) is the better solution for the first week, we must look at the evolution of the displacements over time and determine whether the dike simply opened with fixed length and height, or whether it simultaneously grew vertically and continued to open. In both cases slip on the fault is required to fit the station on Shikinejima.

5.6.3 Evolution

Figure 54 shows the observed (Figure 54A) and modeled dike-normal displacements for three different scenarios. In the first (Figure 54B), we assume that

139 the dike reached its final height and depth during the first week (W-A, depth = 13 km, height ≈ 11 km, opening ≈ 0.2 m) and then opened at constant rate to the preferred model for the entire event (Model F-B, depth = 13 km, height ≈ 11 km, opening ≈ 3 m). In the second, (Figure 54C), the dike grows at constant rate in both vertical extent and in opening from Model W-B after the first week (very shallow) to Model F-

B after the full event. In the third (Figure 54D) we test whether the dike could have grown vertically both upwards and downwards from a 2-km-high dike with base at 5 km depth (Model W-C). For all, fault slip is assumed to increase at constant rate from the initial value to the final value.

Comparing figures for all three scenarios to the observed evolution (Figure

54A), we see that evolution from Models W-B and W-C to Model F-B match the pattern of dike evolution substantially better than does evolution from Model W-A, suggesting that the dike grew in two-dimensions over the course of the event. The fit in 23D is very similar to that shown in Figure 54C. It should be noted that evolution from Model W-A assumes that the dike opens linearly with time, resulting in a series of identical, overlapping dike-normal profiles. Time-varying opening, as would be expected if the dike pressure evolved over time, would yield a different evolution of displacement profiles. Nonetheless, evolution via dike opening only (whether steady or not) reflects a pressure increase from 0.3 MPa after the first week to ~ 6 MPa at the end of the full event. Dike growth in two dimensions (Figure 54C and Figure 54D), on the other hand, reflects a decrease in overpressure from ~16 MPa after the first week to ~6 MPa after the full event. A decrease in overpressure makes sense if the source of magma was initially over-pressured and then underwent a pressure decrease

140 as magma was expelled, making this model of evolution preferable to pure opening.

Given the large uncertainty of models for the first week of activity, however, it is currently not possible to determine exactly how shallow the dike was during the first week of activity. In other words it is not possible to distinguish between a 2-km-high dike at 2 km depth and a 2-km-high dike at 5 km depth based solely on the displacements from the first week of the event. Given that the magma chamber beneath Miyakejima is assumed to be located at 4.6 km depth, the latter (Model W-C) is more likely.

5.7 Discussion

Our preferred model of dike evolution based on GPS data inversion indicates that the dike was initially shallow (2 to 5 km deep) and then grew deeper over time.

This downward propagation is the opposite of what one expects, given that magma is typically less dense than wallrock and thus buoyant. One possibility is that the stress state in the plate is such that the least compressive stress increases towards the surface because of complex plate interactions, causing the dike to propagate downwards.

Wright and Sakai (2005) proposed an alternative explanation, citing the initial deepening of the seismic swarm, and suggested that the propagation of a dike, sourced initially from a magma chamber beneath Miyakejima, led to downward “cracking” of the Philippine Sea Plate. They further hypothesize, consistent with the conclusions of many previous studies, that this downward “cracking” triggered deeper magma sources between Kozushima and Miyakejima that then fed the growing intrusion. We propose a modification of this hypothesis. A series of N-S striking rifts in the Izu backarc has propagated northward over the past ~ 2 Ma, ending with the Hachijo Rift

141 which lies ~70 km south of Miyakejima. We thus propose that the Miyakejima intrusive event may represent the next stage in the northward propagation of the Izu

Arc Rift, which trends northwestward in this region because of stresses imposed by collision of the Philippine Sea Plate with the surrounding plates. The plate is essentially being bowed by subduction to the east and northwest while colliding with the Japanese mainland to the north (inset Figure 32).

5.8 Conclusions

Work presented here allows us to propose answers to the three questions posed in the introduction.

1) Why is there so little deformation during the first week, when the dike appears

to have reached near its full length?

Based on inversion of displacements from the first week and forward modeling of deformation at stations perpendicular to the dike, we propose that the initial dike was quite limited in height (and at least modestly shallow) and then continued to grow vertically and open over time. Forcing the top of the dike to ~10 km (coinciding with the JMA dataset), either for the first week or for the full event results in order of magnitude increases in data misfit. The shallow depth of the best-fitting model is consistent with the small submarine eruption off the coast of Miyakejima after the first day, but seems to conflict with the apparent deepening of the seismic swarm during the first week of activity in the JMA locations. Given that ocean-bottom seismometers were not installed until July 1, it is likely that the depths during the first week may be artificially deep because of poor seismometer geometry.

142

2) Why did the dike stop propagating laterally, although GPS stations recorded

dramatic deformation for two more months?

Pre-event seismicity and the relocated hypocenters during the event show a structure beneath Kozushima and Niijima. Focal mechanisms in this region show consistent left-lateral strike-slip faulting in a direction perpendicular to the assumed dike azimuth. If a fault zone exists between Kozushima and Niijima, it seems logical that the dike may have stopped propagating laterally when it intersected the fault because of differences in material properties, a change in stress regime, or blunting of the dike tip as it intersects the fault zone. Coulomb stress modeling indicates that left-lateral slip on a fault with this orientation would be anticipated if it was intersected by the dike tip. The fault may then have accommodated the dike-normal strain via strike-slip faulting, thus terminating further dike propagation.

3) Where was the magma in relation to the seismicity (both the relocated JMA

catalogue and the ERI dataset)? Is the dike collocated with the plane of

seismicity in either data set, or does it extend beyond?

Restricting the dike to be collocated with the seismicity from the JMA catalogue, results in an underestimate of near-field displacements and an overestimate of far-field displacements. This pattern is indicative of a dislocation model that is too deep, resulting in deformation that has a longer wavelength than the data. The best fitting models for the full event (Models F-B and F-C) coincide with the cloud of seismicity recognized in the ERI hypocenter data. The dike height in these models extends into the zone of diffuse seismicity at 7-2 km depth seen in the ERI hypocenters. The fact that allowing dike depth to vary as well as the height (Model F-C), fits the data only

143 marginally better than fixing the depth to 13 km (Model F-B) suggests that the base of the dike coincides with the cutoff in seismicity in the ERI data. As in previous studies we find a gross difference in dike volume (~ 1 km3) as compared to Mogi source volume (~0.1 km3), suggesting that magma is indeed being supplied from somewhere other than 4 km beneath Miyakejima. Magma compressibility could explain at least part of this discrepancy (Rivalta and Segall, 2008), if the magma expanded after leaving the chamber.

144

5.9 Tables

Table 8: INVERSION RESULTS

Model F-A Model F-B Model F-C Model F-D Model W-A Model W-B

Time Frame Whole event Whole event Whole event Whole event First week First week

Dike parameters

Length (km) 26 (0.54) 30 (0.52) 30 (0.51) 25 (0.21) 30 (2.9) 30 (2.8)

Height (km) 10 11 (0.51) 12 (2.8) 2.7 (0.46) 11 1.0 (1.1)

Depth (km) 20 13 14 (2.0) 7.3 (0.64) 13 1.5 (2.8)

Dip (deg.) 90 90 90 90 90 90

Strike (deg.) 129 129 129 7 129 129

Center (km east, north) 11, -20 10, -18 10, -18 12, -20 10, -18 10, -18

Opening (m) 6.5 (0.42) 3.4 (0.38) 3.1 (1.0) 14 (1.5) 0.18 (0.16) 0.82 (0.87)

Fault parameters

Length (km) 2.2 (0.24) 5.7 (1.4) 5.7 (1.2) 16 (0.82) 5.7 5.7

Height (km) 7.3 (1.7) 7.4 (3.7) 8.0 (3.9) 2.2 (0.98) 7.4 7.4

Depth (km) 7.3 (1.6) 8.4 (3.3) 8.8 (3.5) 7.3 (0.91) 8.4 8.4

Dip (deg.) 90 90 90 90 90 90

Strike (deg.) 36 36 36 36 36 36

Center (km east, north) 2, -10 0, -7 0, -6.5 -4, 12 0, -7 0, -7

Strike slip (m) 12 (1.3) 4.4 (3.6) 4.1 (2.1) 12 (2.0) 0.40 (0.10) 0.39 (0.01)

Mogi parameters

Location (km east, 29, -28 29, -28 29, -28 29, 28 29, -28 29, -28

north)

Depth (km) 4.6 4.6 4.6 4.6 4.6 4.6

Volume (km3 -0.12 (0.01) -0.12 (0.01) -0.12 (0.01) -0.13 (0.001) -0.02 -0.02

σ2 from initial MCMC 329 113 114 113 38 31

Dike overpressure (MPa) 13 6.2 5.4 100 0.33 16

Fault stress drop (MPa) 100 15 14 100 1.5 1.3

Dike volume (km3) 1.7 1.0 1.1 0.90 0.06 0.03

Weighted-r2 7.0e4 2.4e4 2.4e4 2.4e4 384 336

Notes - Results from weighted simplex search algorithm, where estimated parameters are in bold. Center and depth coordinates refer to the center of the bottom edge of the dislocation. Model uncertainty, given in parenthesis, is the standard deviation of each marginal distribution of the posterior probability distribution generated by the MCMC algorithm. Table 8: Miyakejima - Inversion results

145

5.10 Figures

Figure 32: Location of Miyakejima and other Izu Islands after Murase et al. (2005). Small inset map shows regional setting in terms of latitude and longitude. Large map shows locations of Izu Islands in a local reference frame with origin 139.211 E, 34.333 N.

146

Figure 33: JMA epicenter locations, June 26 - August 31. First 6 days of activity are coded by date (color) and all others are grey. Focal mechanisms are shown for the M>5.6 earthquakes.

147

Figure 34: Change in baselines between Niijima and Kozushima before smoothing (green) and after smoothing (blue) after Toda et al. (2002). Black dashed lines represent the dates of the five largest earthquakes. Red dashed lines represent dates of eruptions on Miyakejima. Inset map shows station locations and earthquakes with M>4.0. Discontinuity on August 3, 2000 most likely related to a M 5.3 earthquake near Kozushima. Pink line marks the end of the first week when the dike has seismic swarm stops propagating laterally.

148

Figure 35: Locations of (A) GEONET continuous GPS stations that were used in inversion of deformation; (B) JMA seismic stations used in JMA catalogue relocation; (C) ERI seismic stations used to locate earthquakes from seismic swarm (Sakai et al., 2001). Yellow = on-land seismometers, blue = ocean-bottom seismometers deployed on July 1, 2000.

149

Figure 36: Annual background motion calculated from displacements between June 1997 - June 15 2000, relative to station 0241.

150

Figure 37: (A) Hypocenters before relocation from JMA catalogue. (B) Hypocenters after relocation. M = 6.0 earthquakes are circled in red.

151

Figure 38: Residual distances between originally relocated hypocenters and 100 relocated bootstrap samples for the first 1000 events of the swarm.

152

Figure 39: NIED calculated focal mechanisms for JMA catalogue. (A) Beach-ball representations of focal mechanisms for M>5.0 earthquakes; (B) T-axes and P-axes for earthquakes with M>4.0; (C) Histograms of stress-axes plunges.

153

Figure 40: Pre-event seismicity, 1990-2000, color-coded by magnitude. Focal mechanisms for the largest earthquakes are also shown.

154

Figure 41: ERI seismic data, relocated by Sakai et al. (2001). Includes data from ocean bottom seismometers deployed on July 1, 2000. Only earthquakes with M>3.0 are shown.

155

Figure 42: Horizontal deformation, Jun. 26 - Aug. 31. Observed (blue), calculated (red), where the model includes a Mogi source (fixed) and a vertical dike (green) only. Miyakejima stations were not used in the inversion. With this simple model there is a significant discrepancy between the observed and calculated displacements at the Shikinejima station, as well as poor estimation of the displacements at the Kozushima station.

156

Figure 43: Stress modeling via Coulomb 2.0. In each node, stress from an opening dike is resolved onto a vertical, left-lateral fault with strike perpendicular to the dike. Coulomb stress change is calculated at each node, where red values indicate positive Coulomb stress changes (regions that are likely to fail) whereas blue values indicate negative Coulomb stress changes (regions that are unlikely to fail).

157

Figure 44: Stations at which dike-normal displacements are calculated for Figures 45-48 and 54.

158

Figure 45: Observed (blue) and calculated (red) displacements for the full event - June 26 - August 31, for Model F-A. Dike height and depth are fixed based on the location of the seismic plane observed in the JMA data. Marginal distributions from the MCMC algorithm are shown for varying parameters except Mogi volume. Displacements at Miyakejima stations are not shown for clarity.

159

Figure 46: Observed (blue) and calculated (red) displacements for the full event - June 26 - August 31, for Model F-B. Dike depth is fixed based on the location of the seismic plane observed in the ERI data. Marginal distributions from the MCMC algorithm are shown for varying parameters except Mogi volume. Displacements at Miyakejima stations are not shown for clarity.

160

Figure 47: Observed (blue) and calculated (red) displacements for the full event - June 26 - August 31, for Model F-C. Dike height and depth are both allowed to vary. Marginal distributions from the MCMC algorithm are shown for varying parameters except Mogi volume. Miyake stations removed for clarity.

161

Figure 48: Observed (blue) and calculated (red) displacements for the full event - June 26 - August 31, for Model F-D, which includes a fault striking oblique to the presumed location of the dike. Dike height and depth are both allowed to vary. Marginal distributions from the MCMC algorithm are shown for varying parameters except Mogi volume. Miyake stations removed for clarity.

162

Figure 49: Conceptual representation. Prevent-seismicity (blue), seismic swarm (black), deformation sources (green). Red arrows show possible interaction of dike with the fault zone, in which dike- normal stresses near the fault tip are accommodated by left-lateral slip on a pre-existing fault, potentially causing the dike to stop propagating laterally.

163

Figure 50: Observed (blue) and calculated (red) displacements for the first week of the event - June 26 – July 2, assuming no aseismic slip within the fault zone between Niijima and Kozushima. The poor fit may indicate that seismic or aseismic activity within the fault zone and/or due to the propagating dike significantly affected displacements during the first week.

164

Figure 51: Observed (blue) and calculated (red) displacements for the first week of the event - June 26 – July 2, for Model W-A. Dike height and depth are both fixed to the preferred solution, F-B, as are all fault parameters except slip. Marginal distributions from the MCMC algorithm are shown for varying parameters. Displacements at Miyakejima stations removed for clarity

165

Figure 52: Observed (blue) and calculated (red) displacements for the first week of the event - June 26 – July 2, for Model W-B. Dike height and depth can vary but all fault parameters except slip are fixed to values in preferred solution F-B. Marginal distributions from the MCMC algorithm are shown for varying parameters.

166

Figure 53: Alternative models based on MCMC output from W-B (Figure 52). (A) Model W-C: Dike extends from 5 to 3-km-depth, length is 30 km, with 0.60 m of opening, (X2 = 350); (B) Model W-D: Dike extends from 11 to 9-km-depth, length is 30 km, with 1.4 m of opening, (X2 = 495).

167

Figure 54: Evolution models. Dike-normal displacements (A) Observed; (B) Calculated for dike and fault evolution from first week Model W-A to full event model F-B, allowing dike opening only; (C) Calculated for dike and fault evolution from first week model W-B to full event Model F-B i.e. allowing both downward dike propagation from 1.5 km depth to 13 km depth, and opening over time; (D) Calculated for dike and fault evolution from first week Model W-C to full event Model F-B, in which the dike grows vertically upwards and downwards and opens over time. Although it is evident that the dike grew both vertically and opened during the course of the event, the initial depth of the dike is not well-constrained.

168

6 References

Annen, C., Blundy, J.D., and Sparks, R.S.J., 2006, The genesis of intermediate and

silicic magmas in deep crustal hot zones: Journal of Petrology v. 47, p. 505–539.

Annen, C., and Sparks, R.S.J., 2002, Effects of repetitive emplacement of basaltic

intrusions on thermal evolution and melt generation in the crust: Earth and

Planetary Science Letters v. 203, p. 937–955.

Bachmann, O., and Bergantz, G.W., 2004, On the origin of crystal-poor rhyolites:

Extracted from batholithic crystal mushes: Journal of Petrology v. 45, p. 1565–

1582.

Baker, P.E., 1968, Comparative volcanology and petrology of the Atlantic island arcs:

Bulletin of Volcanology v. 32, p. 189.

Battaglia, M., Williams, M.J., Venezky, D.Y., Hill, D.P., Langbein, J.O., Farrar, C.D.,

Howle, J.F., Sneed, M., and Segall, P., 2003, The Long Valley Caldera GIS

Database: USGS Digital Data Series DDS-81.

Bellon, H., Maury, R.C., Soeria-Atmadja, R., Cotten, J., and Polve, M., 2004, 65 my-

long magmatic activity in Sumatra (Indonesia), from Paleocene to Present:

Bulletin de la Societe Geologique de France v. 175, p. 61–72.

Bohrson, W.A., and Reid, M.R., 1997, Genesis of silicic peralkaline volcanic rocks in

an ocean island setting by crustal melting and open-system processes: Socorro

Island, Mexico: Journal of Petrology v. 38, p. 1137.

Campos-Enriquez, J.O., Dominguez-Mendez, F., Lozada-Zumaeta, A., Morales-

Rodriguez, H.F., and Andaverde-Arredondo, J.A., 2005, Application of the Gauss

169

theorem to the study of silicic calderas: The calderas of La Primavera, Los

Azufres, and Los Humeros (Mexico): Journal of Volcanology and Geothermal

Research v. 147, p. 39–67.

Cann, J.R., 1970, Upward movement of granite magma: Geological Magazine v. 107,

p. 335–340.

Carr, M.J., and Stoiber, R.E., 1990, Volcanism, in The Geology of North America: The

Caribbean Region, Dengo, G., and Case, E.J., ed., p. 375-391.

Charlier, B.L.A., Wilson, C.J.N., Lowenstern, J.B., Blake, S., Van, C., PW, and

Davidson, J.P., 2005, Magma generation at a large, hyperactive silicic volcano

(Taupo, New Zealand) revealed by U-Th and U-Pb systematics in zircons: Journal

of Petrology v. 46, p. 3–32.

Chinnery, M.A., 1964, The strength of the earth’s crust under horizontal shear stress:

Journal of Geophysical Research v. 69, p. 2085–2089.

Christensen, N.I., and Mooney, W.D., 1995, Seismic velocity structure and

composition of the continental crust; a global view: Journal of Geophysical

Research, B v. 100, p. 9761–9788.

Clemens, J.D., and Wall, V.J., 1984, Origin and evolution of a peraluminous silicic

ignimbrite suite: the Violet Town Volcanics: Contributions to Mineralogy and

Petrology v. 88, p. 354–371.

Clift, P., and Vannucchi, P., 2004, Controls on tectonic accretion versus erosion in

subduction zones: Implications for the origin and recycling of the continental

crust: Reviews of Geophysics v. 42, p. 2001–2032.

170

Conder, J.A., Wiens, D.A., and Morris, J., 2002, On the decompression melting

structure at volcanic arcs and back-arc spreading centers: Geophysical Research

Letters v. 29, p. 17.1–17.4.

Cruciani, C., Carminati, E., and Doglioni, C., 2005, Slab dip vs. lithosphere age: No

direct function: Earth and Planetary Science Letters v. 238, p. 298–310. de Bremond d’Ars, J., Jaupart, C., and Sparks, R.S.J., 1995, Distribution of volcanoes

in active margins: Journal of geophysical research v. 100, p. 20421–20432. de Silva, S.L., and Gosnold, W.D., 2007, Episodic construction of batholiths: Insights

from the spatiotemporal development of an ignimbrite flare-up: Journal of

Volcanology and Geothermal Research v. 167, p. 320–335.

Dieterich, J., 1994, A constitutive law for rate of earthquake production and its

application to earthquake clustering: Journal of Geophysical Research v. 99, p.

2601.

Dixon-Spulber, S.D., and Rutherford, M.J., 1983, The origin of rhyolite and

plagiogranite in oceanic crust: an experimental study: Journal of Petrology v. 24,

p. 1-25.

Ducea, M.N., and Barton, M.D., 2007, Igniting flare-up events in Cordilleran arcs:

Geology v. 35, p. 1047–1050.

Erlich, E.N., 1986, Geology of the calderas of Kamchatka and Kurile Islands with

comparison to calderas of Japan and the Aleutians, Alaska: US Geological Survey

Open File Report 86-0291, 300 p.

171

Farrar, C.D., Sorey, M.L., Evans, W.C., Howle, J.F., Kerr, B.D., Kennedy, B.M.,

King, C.Y., and Southon, J.R., 1995, Forest-killing diffuse CO2 emission at

Mammoth Mountain as a sign of magmatic unrest: Nature v. 376, p. 675.

Fiske, R.S., and Tobisch, O.T., 1994, Middle Cretaceous ash-flow tuff and caldera-

collapse deposit in the Minarets Caldera, east-central Sierra Nevada, California:

Geological Society of America Bulletin v. 106, p. 582.

Fliedner, M.M., and Klemperer, S.L., 2000, Crustal structure transition from oceanic

arc to continental arc, eastern Aleutian Islands and Alaska Peninsula: Earth and

Planetary Science Letters v. 179, p. 567–579.

Förster, H.J., Tischendorf, G., and Trumbull, R.B., 1997, An evaluation of the Rb vs.

(Y+ Nb) discrimination diagram to infer tectonic setting of silicic igneous rocks:

Lithos v. 40, p. 261–293.

Foulger, G.R., Julian, B.R., Pitt, A.M., Hill, D.P., Malin, P.E., and Shalev, E., 2003,

Three-dimensional crustal structure of Long Valley Caldera, California, and

evidence for the migration of CO2 under Mammoth Mountain: Journal of

Geophysical Research, v. 108, p. 2147-2162.

Fujita, E., Ukawa, M., Yamamoto, E., and Okada, Y., 2002, Cyclic jerky opening of

magma sheet and caldera formation during the 2000 Miyakejima Volcano

eruption: Geophysical Research Letters v. 29, p. 1326-1329.

Furuya, M., Okubo, S., Kimata, F., Miyajima, R., Meilano, I., Sun, W., Tanaka, Y.,

and Miyazaki, T., 2003, Mass budget of the magma flow in the 2000 volcano-

seismic activity at Izu-islands, Japan: Earth, Planets and Space v. 55, p. 375-385.

172

Geyer, A., and Martì, J., 2008, The new worldwide collapse caldera database (CCDB):

A tool for studying and understanding caldera processes: Journal of Volcanology

and Geothermal Research v. 175, p. 334–354.

Grebennikov, A.V., and Maksimov, S.O., 2006, Fayalite rhyolites and a zoned magma

chamber of the Paleocene Yakutinskaya volcanic depression in Primorye, Russia:

Journal of Mineralogical and Petrological Sciences v. 101, p. 69–88.

Greene, D.C., and Schweickert, R.A., 1995, The Gem Lake shear zone: Cretaceous

dextral transpression in the northern Ritter Range pendant, eastern Sierra Nevada,

California: Tectonics v. 14, p. 945–945.

Gripp, A.E., and Gordon, R.G., 2002, Young tracks of hotspots and current plate

velocities: Geophysical Journal International v. 150, p. 321–361.

Heidbach, O., Tingay, M., Barth, A., Reinecker, J., Kurfess, D., and Muller, B., 2008,

The 2008 release of the World Stress Map: www.world-stress-map.org.

Hildreth, W., and Moorbath, S., 1988, Crustal contributions to arc magmatism in the

Andes of Central Chile: Contributions to Mineralogy and Petrology v. 98, p. 455–

489.

Hildreth, W., Fierstein, J., Godoy, E., Drake, R.E., and Singer, B., 1999, The Puelche

Volcanic Field: Extensive Pleistocene rhyolite lava flows in the Andes of central

Chile: Revista Geológica de Chile v. 26, p. 275–309.

Hildreth, W., 1981, Gradients in silicic magma chambers; implications for lithospheric

magmatism: Journal of Geophysical Research, B v. 86, p. 10153–10192.

173

Hildreth, W., 2004, Volcanological perspectives on Long Valley, Mammoth

Mountain, and Mono Craters: several contiguous but discrete systems: Journal of

Volcanology and Geothermal Research v. 136, p. 169–198.

Hill, D.P., Ellsworth, W.L., Johnston, M.J.S., Langbein, J.O., Oppenheimer, D.H.,

Pitt, A.M., Reasenberg, P.A., Sorey, M.L., and McNutt, S.R., 1990, The 1989

earthquake swarm beneath Mammoth Mountain, California: An initial look at the

4 May through 30 September activity: Bulletin of the Seismological Society of

America v. 80, p. 325–339.

Hill, D.P., 1992, Temperatures at the Base of the Seismogenic Crust beneath Long

Valley Caldera, California and the Phlegrean Fields Caldera, Italy: Volcanic

Seismology v. 3, p. 432-461.

Hill, D.P., and Prejean, S., 2005, Magmatic unrest beneath Mammoth Mountain,

California: Journal of Volcanology and Geothermal Research v. 146, p. 257.

Hochstein, M.P., 1995, Crustal heat transfer in the Taupo Volcanic Zone (New

Zealand): Comparison with other volcanic arcs and explanatory heat source

models: Journal of Volcanology and Geothermal Research v. 68, p. 117–151.

Houghton, B.F., Wilson, C.J.N., McWilliams, M.O., Lanphere, M.A., Weaver, S.D.,

Briggs, R.M., and Pringle, M.S., 1995, Chronology and dynamics of a large silicic

magmatic system; central Taupo volcanic zone, New Zealand: Geology v. 23, p.

13-16.

Hughes, G.R., and Mahood, G.A., 2008, Tectonic controls on the nature of large

silicic calderas in volcanic arcs: Geology v. 36, p. 627-630.

174

Humphreys, E.D., 1995, Post-Laramide removal of the Farallon slab, western United

States: Geology v. 23, p. 987-990.

Irwan, M., Kimata, F., and Fujii, N., 2006, Time dependent modeling of magma

intrusion during the early stage of the 2000 Miyakejima activity: Journal of

Volcanology and Geothermal Research v. 150, p. 202–212.

Irwan, M., Kimata, F., Fujii, N., Nakao, S., Watanabe, H., Sakai, S., Ukawa, M.,

Fujita, E., and Kawai, K., 2003, Rapid ground deformation of the Miyakejima

Volcano on 26-27 June 2000 detected by kinematic GPS analysis: Earth, Planets

and Space v. 55, p. 13-16.

Ito, T., and Yoshioka, S., 2002, A dike intrusion model in and around Miyakejima,

Niijima and Kozushima in 2000: Tectonophysics v. 359, p. 171-187.

Jellinek, A.M., and DePaolo, D.J., 2003, A model for the origin of large silicic magma

chambers: precursors of caldera-forming eruptions: Bulletin of Volcanology v. 65,

p. 363–381.

John, D.A., Henry, C.D., and Colgan, J.P., 2008, Magmatic and tectonic evolution of

the Caetano caldera, north-central Nevada: A tilted, mid-Tertiary eruptive center

and source of the Caetano Tuff: Geosphere v. 4, p. 75-106.

Kay, S.M., and Coira, B.L., 2009, Shallowing and steepening subduction zones,

continental lithospheric loss, magmatism, and crustalflow under the Central

Andean Altiplano-Puna Plateau: Backbone of the Americas: shallow subduction,

plateau uplift, and ridge and terrane collision, p. 229.

175

Kay, S.M., Ramos, V.A., Mpodozis, C., and Sruoga, P., 1989, Late Paleozoic to

Jurassic silicic magmatism at the Gondwana margin: Analogy to the Middle

Proterozoic in North America?: Geology v. 17, p. 324-328.

Kemp, A.I.S., Hawkesworth, C.J., Paterson, B.A., Foster, G.L., Kinny, P.D.,

Whitehouse, M.J., and Maas, R., 2008, Exploring the plutonic--volcanic link: a

zircon U--Pb, Lu--Hf and O isotope study of paired volcanic and granitic units

from southeastern Australia: Earth and Environmental Science Transactions of the

Royal Society of Edinburgh v. 97, p. 337–355.

Kitanidis, P.K., 1996, On the geostatistical approach to the inverse problem: Advances

in Water Resources v. 19, p. 333–342.

Lagarias, J.C., Reeds, J.A., Wright, M.H., and Wright, P.E., 1998, Convergence

Properties of the Nelder--Mead Simplex Method in Low Dimensions: SIAM

Journal on Optimization v. 9, p. 112–147.

Lallemand, S., Heuret, A., and Boutelier, D., 2005, On the relationships between slab

dip, back-arc stress, upper plate absolute motion, and crustal nature in subduction

zones: Geochemistry, Geophysics, Geosystems v. 6, 18 p.

Langbein, J.O., 2003, Deformation of the Long Valley Caldera, California: inferences

from measurements from 1988 to 2001: Journal of Volcanology and Geothermal

Research v. 127, p. 247–267.

Langbein, J., Dzurisin, D., Marshall, G., Stein, R., and Rundle, J., 1995, Shallow and

peripheral volcanic sources of inflation revealed by modeling two-color

geodimeter and leveling data from Long Valley Caldera, California, 1988-1992:

Journal of Geophysical Research, v. 100, p. 12487-12496.

176

Langbein, J.O., Hill, D.P., Parker, T.N., and Wilkinson, S.K., 1993, An episode of

reinflation of the Long Valley Caldera, eastern California; 1989-1991: Journal of

Geophysical Research, v. 98, p. 15851-15870.

Lee, C.T.A., 2005, Trace element evidence for hydrous metasomatism at the base of

the North American lithosphere and possible association with Laramide low-angle

subduction: The Journal of Geology v. 113, p. 673–685.

Lowe, T., 1996, Petrogenesis of the Minarets and Merced Peak volcanic-plutonic

complexes, Sierra Nevada, California: Geological and Environmental Sciences v.

PhD, p. 157.

Lowenstern, J.B., and Mahood, G.A., 1991, New data on magmatic H 2 O contents of

pantellerites, with implications for petrogenesis and eruptive dynamics at

Pantelleria: Bulletin of Volcanology v. 54, p. 78–83.

Mahood, G.A., 1990, Second reply to comment of RSJ Sparks, HE Huppert and CJN

Wilson on Evidence for long residence times of rhyolitic magma in the Long

Valley magmatic system: The isotopic record in the precaldera lavas of Glass

Mountain: Earth and Planetary Science Letters v. 99, p. 395–399.

Mahood, G.A., and Halliday, A.N., 1988, Generation of high-silica rhyolite: A Nd, Sr,

and O isotopic study of Sierra La Primavera, Mexican Neovolcanic Belt:

Contributions to Mineralogy and Petrology v. 100, p. 183–191.

Mahood, G.A., Ring, J.H., Manganelli, S., and McWilliams, M.O., 2010, New

40Ar/39Ar ages reveal contemporaneous mafic and silicic eruptions during the past

160,000 years at Mammoth Mountain and Long Valley caldera, California:

Geological Society of America Bulletin v. 122, p. 396-407.

177

Miller, C.F., and Wark, D.A., 2008, Supervolcanoes and their explosive

supereruptions: Elements v. 4, p. 11-16.

Miyazaki, S., Segall, P., McGuire, J.J., Kato, T., and Hatanaka, Y., 2006, Spatial and

temporal evolution of stress and slip rate during the 2000 Tokai slow earthquake:

Journal of Geophysical Research, v. 111, 3409-3425.

Mosegaard, K., and Tarantola, A., 1995, Monte Carlo sampling of solutions to inverse

problems: Journal of Geophysical Research, B v. 100, p. 12–431.

Murase, M., Irwan, M., Kariya, S., Tabei, T., Okuda, T., Myajima, R., Oikawa, J.,

Watanabe, H., Kato, T., Nakao, S., Ukawa, M., Fujita, E., Okayama, M., Kimata,

F., and Fujii, N., 2006, Time dependent model of magma intrusion in and around

Miyake and Kozu Islands, Central Japan in June-August, 2000: Journal of

Volcanology and Geothermal Research v. 150, p. 213–231.

Murray, J., 2003, Spatial and temporal slip-rate variations on the San Andreas fault

inferred from geodetic data and the implications for strain accumulation: PhD

Dissertation, Dept. of Geophysics, Stanford University, p. 177.

Newhall, C.G., and Dzurisin, D., 1988, Historical unrest at large calderas of the world,

Volumes 1 and 2: US Geological Survey Bulletin v. 1855, p. 1108.

Nishimura, T., Ozawa, S., Murakami, M., Sagiya, T., Tada, T., Kaidzu, M., and

Ukawa, M., 2001, Crustal deformation caused by magma migration in the northern

Izu Islands, Japan: Geophysical Research Letters v. 28, p. 3745-3748.

Oppenheimer, C., 2003, Climatic, environmental and human consequences of the

largest known historic eruption; Tambora Volcano (Indonesia) 1815: Progress in

Physical Geography v. 27, p. 230-259.

178

Ozawa, S., Miyazaki, S., Nishimura, T., Murakami, M., Kaidzu, M., Imakiire, T., and

Ji, X., 2004, Creep, dike intrusion, and magma chamber deflation model for the

2000 Miyake eruption and the Izu islands earthquakes: Journal of Geophysical

Research, v. 109, p. 2410-2421.

Patino Douce, A.E., 1999, What do experiments tell us about the relative contributions

of crust and mantle to the origin of granitic magmas?: Geological Society London

Special Publications v. 168, p. 55-75.

Pearce, J., 1995, Tectonic implications of the composition of volcanic arc magmas:

Annual Review of Earth and Planetary Sciences v. 23, p. 251–285.

Pearce, J.A., Harris, N.B.W., and Tindle, A.G., 1984, Trace element discrimination

diagrams for the tectonic interpretation of granitic rocks: Journal of Petrology v.

25, p. 956–983.

Petrinovic, I.A., Riller, U., and Brod, J.A., 2005, The Negra Muerta Volcanic

Complex, southern Central Andes: geochemical characteristics and magmatic

evolution of an episodically active volcanic centre: Journal of Volcanology and

Geothermal Research v. 140, p. 295–320.

Pope, D.C., and Willett, S.D., 1998, Thermal-mechanical model for crustal thickening

in the central Andes driven by ablative subduction: Geology v. 26, p. 511–514.

Prejean, S., Stork, A., Ellsworth, W., Hill, D., and Julian, B., 2003, High precision

earthquake locations reveal seismogenic structure beneath Mammoth Mountain,

California: Geophysical Research Letters v. 30, p. 2247-2250.

Rish, I., 2001, An empirical study of the naive Bayes classifier, IJCAI 2001 Workshop

on Empirical Methods in Artificial Intelligence: p. 41–46.

179

Rivalta, E., and Segall, P., 2008, Magma compressibility and the missing source for

some dike intrusions: Geophysical Research Letters v. 35, p. 4306-4310.

Rubin, A.M., 1995, Propagation of magma-filled cracks: Annual Review of Earth and

Planetary Sciences v. 23, p. 287-336.

Rubin, A.M., Gillard, D., and Got, J.L., 1998, A reinterpretation of seismicity

associated with the January 1983 dike intrusion at Kilauea Volcano, Hawaii:

Journal of Geophysical Research v. 103, p. 10003-10016.

Rudnick, R.L., 1995, Making continental crust: Nature (London) v. 378, p. 571–578.

Saito, G., Stimac, J.A., Kawanabe, Y., and Goff, F., 2002, Mafic-felsic magma

interaction at Satsuma-Iwojima volcano, Japan: Evidence from mafic inclusions in

rhyolites: Earth Planets and Space v. 54, p. 303–326.

Sakai, S., Yamada, T., Ide, S., Mochizuki, M.,ShiobaraHajime, Urabe, T., Hirata, N.,

Shinohara, M., Kanazawa, T., Nishizawa, A., Fujie, G., and Mikada., Magma

Migration from the Point of View of Seismic Activity in the Volcanism of

Miyake-jima Island in 2000: Journal of Geography v. 110, p. 145–155.

Scaillet, B., and Macdonald, R., 2003, Experimental constraints on the relationships

between peralkaline rhyolites of the Kenya Rift Valley: Journal of Petrology v. 44,

p. 1867.

Schmincke, H.U., 2004, Volcanism: University of Kiel, Department of Volcanology

and Petrology, Springer-Verlag Berlin, Berlin, Federal Republic of Germany

(DEU).

180

Segall, P., Cervelli, P., Owen, S., Lisowski, M., and Miklius, A., 2001, Constraints on

dike propagation from continuous GPS measurements: Journal of Geophysical

Research, v. 106, p. 19301-19317.

Siebert, L., and Simkin, T., 2002, Volcanoes of the World: an Illustrated Catalog of

Holocene Volcanoes and their Eruptions: Smithsonian Institution, Global

Volcanism Program Digital Information Series, GVP-3,

http://www.volcano.si.edu/world/ v. Last accessed February 1, 2009.

Sigmarsson, O., Condomines, M., and Fourcade, S., 1992, A detailed Th, Sr and O

isotope study of Hekla: differentiation processes in an Icelandic volcano:

Contributions to Mineralogy and Petrology v. 112, p. 20–34.

Small, C., and Naumann, T., 2001, The global distribution of human population and

recent volcanism: Global Environmental Change B: Environmental Hazards v. 3,

p. 93–109.

Smith, R.L., 1979, Ash-flow magmatism: GSA Special Papers v. 180, p. 5–27.

Sparks, S., Self, S., Grattan, J., Oppenheimer, C., Pyle, D., and Rymer, H., 2005,

Super-eruptions: global effects and future threats: Report of a Geological Society

of London Working Group.

Takahashi, N., Uyehiro, K.S., and Hinohara, M.S., 1998, Implications from the

seismic crustal structure of the northern Izu-Bonin arc: The Island Arc v. 7, p.

383–394.

Toda, S., Stein, R.S., and Sagiya, T., 2002, Evidence from the AD 2000 Izu Islands

earthquake swarm that stressing rate governs seismicity: Nature v. 419, p. 58-61.

181

Tilling, R.I., 2004, Reducing Risk at Continental Margin Volcanoes: GSA Penrose

Conference: Neogene-Quaternary Continental Margin Volcanism.

Ueda, H., Fujita, E., Ukawa, M., Yamamoto, E., Irwan, M., and Kimata, F., 2005,

Magma intrusion and discharge process at the initial stage of the 2000 activity of

Miyakejima, central Japan, inferred from tilt and GPS data: Geophysical Journal

International v. 161, p. 891-906.

Verma, S.P., 2001, Geochemical and Sr--Nd--Pb isotopic evidence for a combined

assimilation and fractional crystallization process for volcanic rocks from the

Huichapan caldera, Hidalgo, Mexico: Lithos v. 56, p. 141–164.

Waldhauser, F., and Ellsworth, W.L., 2000, A Double-Difference Earthquake

Location Algorithm: Method and Application to the Northern Hayward Fault,

California: Bulletin of the Seismological Society of America v. 90, p. 1353-1368.

Wright, T.L., and Sakai, S., 2005, Interpretation of the Miyakejima 2000 eruption and

dike emplacement using time animations of earthquakes: Bulletin of the

Earthquake Research Institute, University of Tokyo v. 79, p. 1–16.

Yamaoka, K., Kawamura, M., Kimata, F., Fujii, N., and Kudo, T., 2005, Dike

intrusion associated with the 2000 eruption of Miyakejima Volcano, Japan:

Bulletin of Volcanology v. 67, p. 231–242.

182

7 Appendix A

Data for arc calderas used in Chapter 2.

Codes • Crust type: 1 = oceanic, 2 = transitional, 3 = continental • Caldera-forming eruption composition: 1 = dacite, 2 = rhyodacite, 3 = rhyolite • Local stress: -2 = compression, -1 = transpression, 1 = transtension, 2 = extension • Upper plate stresses: -3 = thrusting, -2 = moderate compression, -1 = slight compression, 0 = neutral or strike-slip, 1 = slight extension, 2 = moderate extension, 3 = backarc spreading • Oldest crust code: 1 = oceanic, 2 = Tertiary, 3 = Mesozoic, 4 = Paleozoic, 5 = Precambrian

Notes • Convergence obliquity= Angle between convergence vector and trench-normal vector • Numbers in the reference column contain reference index numbers that correspond to the citations listed in Appendix C • Unless otherwise noted, convergence rates, slab ages, and upper plate stress codes, are taken from Lallemand et al. 2005 (15561) and slab dip and obliquity were taken from Cruciani et al. 2005 (17317).

183

0 7 0 0 0 0 0 0 30 60 25 20 60 65 50 110 250 500 250 109 300 km) Volume (cubic- Volume 7 9 6 8 9 11 18 14 45 12 15 18 15 19 20 20 21 8.5 7.25 12.5 22.5 Average Average Diameter (km) 0 4 0 0 1 0 9 0 2 0 5 0 0 0 12 14 32 87 12 16 16 front (km) Distance from 3 3 3 3 3 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3 3 Crust type 106 121 98.83 131.2 115.13 115.38 131.11 25.396 27.093 103.92 100.18 121.27 124.05 130.71 130.82 156.93 157.53 116.414 105.423 130.308 130.679 Longitude 14 -6.3 2.58 14.4 -4.83 -0.33 -8.28 -8.24 12.77 31.65 32.03 32.88 51.43 -8.408 -6.102 36.404 36.676 30.789 31.345 33.167 51.806 Latitude 1 1 2 2 2 3 3 3 3 3 4 4 5 6 6 6 6 6 6 7 7 Arc number Location Greece Greece Indonesia - Sumatra Indonesia - Sumatra Indonesia - Sumatra Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Philippines N. Philippines N. Philippines S. Japan-Kyushu -Ryukyu Islands Japan-Kyushu Japan-Kyushu Japan-Kyushu Japan-Kyushu Japan-Kyushu arc Kamchatka - Volcanic arc Kamchatka - Volcanic Caldera name Santorini Kos Ranau Maninjau Toba Anak Segara Bratan Batur Dano (Danau or Banten) Krakatau Taal Laguna de Bay Irosin Kikai Ata Aira Kakuto Aso Shishimuta Pauzhetka- Kurile Lake Ksudach Massif

184

6 0 6 0 0 0 0 0 0 0 0 90 20 46 20 80 115 115 120 800 km) Volume (cubic- Volume 5 7 6 8 8 13 15 18 10 10 19 23 16 10 6.5 6.5 8.5 24.5 10.5 11.25 Average Average Diameter (km) 0 0 0 2 2 2 0 0 2 6 11 59 19 14 19 16 26 13 10 10 front (km) Distance from 3 3 3 3 3 3 3 3 3 2 2 1 1 1 1 1 1 1 1 1 Crust type 147 147.2 148.8 158.03 160.02 159.97 144.43 145.74 148.22 150.88 151.95 154.71 157.332 157.968 159.464 159.452 159.593 160.273 144.013 145.501 Longitude 54.5 44.6 52.78 54.32 43.55 43.98 45.38 46.52 46.92 49.36 54.115 52.513 52.558 53.981 54.071 54.593 43.384 43.865 44.625 45.192 Latitude 7 7 7 7 7 7 7 7 7 8 8 8 8 8 8 8 8 8 8 8 Arc number Location Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic Kurile - N. Japan Kurile - N. Japan Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Caldera name Opala Gorely Karymshina Akademii--Polovinka Karymsky Stena-Soboliny Bolshoi Semiachik Uzon-Geyzernaya Krasheninnikov Akan Kutcharo (Mashu) Golovnin Mendeleev L'vinaya Past Urbich Isthmus Vetrovoi Medvezhii or Medvezhia Chirpoi Zavaritsky Tao-Rusyr

185

0 0 0 0 0 0 0 0 0 10 50 74 80 60 40 100 130 150 125 km) Volume (cubic- Volume 6 8 8 6 7 11 10 18 15 10 14 13 25 6.5 8.5 6.5 7.5 6.5 11.5 Average Average Diameter (km) 0 2 0 7 0 7 0 4 0 2 11 14 14 59 55 18 28 12 16 front (km) Distance from 1 3 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 2 Crust type 154.81 140.73 140.77 140.88 141.32 140.82 143.15 142.88 140.05 140.01 139.85 139.92 139.66 150.52 139.845 140.518 140.695 140.881 140.854 Longitude 32.1 33.4 49.57 38.73 38.83 40.65 42.75 43.08 43.68 31.47 31.95 32.45 -5.581 37.194 38.736 40.469 40.579 42.599 43.516 Latitude 8 9 9 9 9 9 9 9 9 9 9 9 9 10 10 10 10 10 12 Arc number Location Kurile Islands Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan-Izu Islands Japan-Izu Islands Japan-Izu Islands Japan-Izu Islands Japan-Izu Islands Papua New Guinea Caldera name Nemo Hatori Narugo Mukaimachi-Akakura Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi-Mitsumata Daisetsu Sumisu Myojinsho Myojin Knoll Higashi-aogashima Kurose Hole Witori-Pago

186

0 5 5 10 35 16 50 50 34 55 10 115 530 100 175 150 225 100 km) 1000 Volume (cubic- Volume 8 9 6 8 9 7 11 12 22 35 12 22 20 43 21 9.5 10.5 14.5 14.5 Average Average Diameter (km) 0 8 8 4 0 3 0 0 8 0 33 16 15 21 53 27 19 57 109 front (km) Distance from 2 2 2 1 1 1 3 3 3 3 3 3 3 1 3 3 3 3 3 Crust type 152.2 168.54 176.33 -162.08 -122.12 151.158 152.203 175.925 176.061 175.782 176.272 176.501 176.244 179.607 -178.431 -177.964 -164.386 -158.145 -121.229 Longitude -4.92 55.34 42.93 -4.271 -4.122 -16.83 -29.27 -38.42 51.944 54.668 56.904 43.722 -30.232 -38.815 -38.553 -38.364 -38.252 -38.182 -38.085 Latitude 12 12 12 13 15 15 16 16 16 16 16 16 16 17 18 18 18 19 19 Arc number Location Papua New Guinea Papua New Guinea Papua New Guinea (New Hebrides) Vanuatu Kermadec Islands Kermadec Islands New Zealand New Zealand New Zealand New Zealand New Zealand New Zealand New Zealand Alaskan Islands (Aleutians) US Alaskan Peninsula - US Alaskan Peninsula - US Alaskan Peninsula - US Cascades - US Cascades - US Caldera name Lolobau Rabaul Tavui Kuwae Macauley Island Denham Taupo Whakamaru-Maroa Reporoa Mangakino Kapenga Okataina Caldera Complex Rotorua Semisopochnoi Fisher Emmons Lake Aniakchak Crater Lake Newberry

187

0 0 0 5 7 1 0 0 0 0 0 33 20 18 25 30 32 90 115 275 100 km) Volume (cubic- Volume 5 6 8 6 11 18 15 20 15 6.3 5.5 9.5 6.5 5.3 5.5 5.25 18.5 18.5 17.5 17.5 26.5 Average Average Diameter (km) 7 8 7 4 5 9 3 6 2 0 10 61 16 16 15 54 70 33 40 89 25 front (km) Distance from 3 3 3 3 3 3 3 3 3 3 2 2 2 2 1 3 3 3 3 3 3 Crust type -97.45 -103.5 -88.93 -89.55 -90.59 -85.33 -86.03 -80.17 -78.39 -68.83 -67.74 -67.75 -121.71 -88.518 -89.053 -89.294 -91.193 -85.153 -86.935 -78.322 -78.436 Longitude -23 8.58 -0.27 -27.9 11.92 48.83 19.68 20.67 13.63 13.87 10.83 -0.787 -0.677 -26.76 13.495 13.672 13.734 14.449 14.678 10.748 12.677 Latitude 19 20 20 21 21 21 21 21 21 21 22 22 22 22 23 24 24 24 25 25 25 Arc number Location Cascades - US Mexico Mexico America N. El Salvador Central America N. El Salvador Central America N. El Salvador Central America N. El Salvador Central America N. El Salvador Central America N. - Guatemala Central America N. - Guatemala Central America S. - Costa Rica Central America S. - Costa Rica Central America S. - Nicaragua Central America S. - Nicaragua Central Panama Andes N. - Ecuador Andes N. Ecuador Andes N. - Ecuador Argentina Andes C. - Argentina Andes C. - Andes C. - Chile Caldera name Kulshan Humeros, Los Primavera, La Berlin Carbonera Caldera, La Ilopongo Boqueron Coatepeque Amatitlan Atitlan Guayabo Guachipelin Apoyo Pelona El Valle Barrancas Chalupas Chacana Incapillo Robledo (Cerro Blanco) Purico Complex

188

0 7 0 20 58 113 350 km) 1000 Volume (cubic- Volume 9 6 12 16 20 18 10 9.5 Average Average Diameter (km) 0 2 2 46 38 53 23 10 front (km) Distance from 3 3 3 3 3 3 1 1 Crust type -72.22 -71.17 -71.06 -69.78 -61.33 -70.507 -70.496 -61.317 Longitude -40.5 15.33 15.37 -38.67 -37.837 -36.063 -35.558 -34.171 Latitude 26 26 26 26 26 26 28 28 Arc number Location Andes S. - Chile Andes S. - Chile Andes S. - Chile Andes S. - Chile Andes S. - Chile Argentina Andes S. - Antilles Lesser Antilles Lesser Caldera name Cordillera Nevada Pino Hachado Copahue-Caviahue Bobadilla Calabozos Maipo-Diamante Waven Wotten Pitons Trois Morne

189

38 19 50 37 28 66 65 66 53 53 41 41 53 73 73 73 73 73 73 77 77 rate (mm/yr) convergence Trench-normal Trench-normal - - 72 47 46 84 83 83 75 75 22 22 45 50 50 50 50 50 50 110 110 Slab age at trench (Ma) 35 35 38 41 35 37 34 36 38 38 40 40 52 45 45 45 45 45 45 38 38 km (degrees) Slab dip at 100 5 8 8 8 5 5 6 6 6 6 3 3 11 11 34 19 39 35 43 43 66 obliquity (degrees) Convergence 2 2 1 1 1 2 2 2 1 1 2 1 2 2 2 2 2 2 1 1 -1 code Local stress 32 27 30 30 30 20 20 20 20 20 34 34 45 30 30 30 30 30 30 30 32 Crustal thickness (km) 0 0 52 37 85 22 50 0.8 1.6 6.3 110 172 161 550 840 140 900 300 250 29.3 1000 CFE age (Ma) - 0.0 70.3 77.0 73.0 74.0 74.4 66.2 68.0 64.0 64.0 69.0 76.0 72.0 74.0 76.5 75.0 66.0 64.9 71.5 71.8 SiO2 Highest wt% 2 3 3 3 3 1 1 1 2 1 1 2 3 2 3 3 3 1 1 3 2 CFE code composition Caldera name Santorini Kos Ranau Maninjau Toba Anak Segara Bratan Batur Dano Krakatau Taal Laguna de Bay Irosin Kikai Ata Aira Kakuto Aso Shishimuta Pauzhetka Ksudach

190

76 76 76 75 75 74 74 74 74 77 77 74 74 75 75 76 71 77 78 78 rate (mm/yr) convergence Trench-normal Trench-normal 119 119 118 118 118 114 110 106 106 103 100 100 100 100 100 100 128 128 120 120 Slab age at trench (Ma) 38 38 38 38 38 38 38 38 38 38 38 38 38 38 38 39 40 40 39 39 km (degrees) Slab dip at 100 2 2 7 8 12 12 17 17 17 17 33 33 31 31 30 30 28 26 19 15 obliquity (degrees) Convergence 2 1 1 2 1 2 1 1 1 2 1 2 -2 -2 -2 -2 -2 -2 -2 -2 code Local stress 37 32 32 37 37 37 37 37 37 28 28 25 25 20 20 21 21 17 17 25 Crustal thickness (km) 0 0 0 0 40 40 40 39 39 7.7 9.4 7.5 180 180 560 340 400 39.5 31.5 1780 CFE age (Ma) 0.0 76.0 69.5 71.9 65.4 70.0 67.0 74.4 73.3 72.4 71.9 70.0 67.0 67.1 67.8 74.0 72.0 72.0 67.0 SiO2 Highest wt% 3 2 3 1 2 1 3 3 1 3 2 2 1 2 1 1 3 2 2 1 CFE code composition Caldera name Opala Gorely Karymshina Akademii Karymsky Stena-Soboliny Bolshoi Sem. Uzon Krasheninnikov Akan Kutcharo Golovnin Mendeleev L'vinaya Past Urbich Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr

191

77 90 86 86 86 88 88 88 82 82 82 77 77 54 54 54 55 56 98 rate (mm/yr) convergence Trench-normal Trench-normal 30 110 129 131 131 131 130 130 130 128 128 128 128 128 131 131 131 129 128 Slab age at trench (Ma) 38 30 30 30 30 32 32 32 36 36 36 38 38 44 44 44 40 38 59 km (degrees) Slab dip at 100 6 8 0 10 10 10 20 20 20 29 29 29 31 31 27 27 27 26 22 obliquity (degrees) Convergence 2 2 2 2 2 2 2 -2 -2 -2 -2 -2 -2 -2 -2 -1 -2 -2 -2 code Local stress 25 30 30 30 30 30 30 30 30 30 30 34 40 23 23 23 23 23 30 Crustal thickness (km) 0 0 0 0 0 73 55 41 30 1.7 3.3 760 130 600 160 1300 1500 2000 2000 CFE age (Ma) 68.0 72.0 75.0 64.0 72.0 72.0 72.0 72.0 76.9 74.0 77.3 71.6 72.0 69.0 66.4 75.0 73.2 68.0 66.2 SiO2 Highest wt% 2 2 3 1 2 2 2 2 3 3 3 2 2 2 1 3 2 1 1 CFE code composition Caldera name Nemo Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll Higashi Kurose Hole Witori-Pago

192

90 56 57 38 38 38 38 38 43 43 42 33 62 59 24 24 120 143 143 rate (mm/yr) convergence Trench-normal Trench-normal 11 11 30 30 30 51 30 30 30 30 30 30 30 56 53 52 52 100 100 Slab age at trench (Ma) 57 57 57 53 55 53 47 47 47 47 47 48 48 63 55 47 34 55 55 km (degrees) Slab dip at 100 0 0 0 2 14 20 15 32 32 32 32 32 30 30 43 13 12 27 28 obliquity (degrees) Convergence 1 1 1 2 2 2 2 2 1 2 1 2 2 2 -2 -2 -2 -2 -2 code Local stress 30 32 32 28 15 15 20 20 20 20 20 20 20 27 33 37 33 40 37 Crustal thickness (km) 0 12 1.4 7.1 0.5 6.3 2.2 1.6 9.1 3.4 6.8 340 240 280 220 238 500 26.5 0.89 CFE age (Ma) 69.0 64.8 75.0 64.5 72.0 67.7 76.0 77.0 75.0 76.0 73.0 76.0 76.0 66.0 67.9 75.0 70.3 72.0 72.4 SiO2 Highest wt% 2 1 3 1 2 1 3 3 3 3 3 3 3 1 1 3 2 2 2 CFE code composition Caldera name Lolobau Rabaul Tavui Kuwae Macauley Denham Taupo Whakamaru Reporoa Mangakino Kapenga Okataina Rotorua Semisopochnoi Fisher Emmons Lake Aniakchak Crater Lake Newberry

193

37 60 44 75 73 73 73 73 68 68 81 81 80 78 14 38 38 33 65 68 72 rate (mm/yr) convergence Trench-normal Trench-normal 8 10 15 25 24 24 24 24 22 22 27 27 27 28 25 18 18 15 52 52 55 Slab age at trench (Ma) 53 25 40 46 45 45 45 45 43 43 49 49 49 47 37 45 45 45 21 20 31 km (degrees) Slab dip at 100 5 0 2 2 2 2 0 0 3 1 9 25 15 16 16 74 32 32 38 26 23 obliquity (degrees) Convergence 2 2 2 2 1 1 1 1 1 2 2 1 2 2 2 2 2 1 1 -2 -1 code Local stress 40 42 26 35 36 36 37 37 47 47 37 37 33 35 22 50 50 50 70 70 70 Crustal thickness (km) 95 40 72 84 23 211 460 100 550 191 665 500 980 510 200 1150 2000 1600 2000 1300 1300 CFE age (Ma) 0.0 0.0 73.3 77.0 77.0 65.4 65.7 69.2 69.9 74.7 75.8 77.0 76.0 66.0 68.0 74.0 73.0 75.0 71.5 65.2 66.4 SiO2 Highest wt% 2 3 3 1 1 2 1 2 3 3 3 3 1 1 1 3 3 3 2 1 1 CFE code composition Caldera name Kulshan Humeros, Los Primavera, La Berlin Carbonera Ilopongo Boqueron Coatepeque Amatitlan Atitlan Guayabo Guachipelin Apoyo Pelona El Valle Barrancas Chalupas Chacana Incapillo Robledo Purico

194

11 11 75 74 74 67 62 70 rate (mm/yr) convergence Trench-normal Trench-normal 25 34 34 37 39 45 98 98 Slab age at trench (Ma) 40 34 34 34 34 28 53 53 km (degrees) Slab dip at 100 12 21 21 29 32 26 26 26 obliquity (degrees) Convergence 1 1 1 1 2 2 2 -2 code Local stress 32 32 32 32 47 55 30 30 Crustal thickness (km) 30 40 132 950 300 450 1600 2000 CFE age (Ma) 0.0 63.0 71.8 74.1 71.0 70.5 74.3 63.0 SiO2 Highest wt% 1 3 3 2 2 3 1 1 CFE code composition Caldera name Cord. Nevada Pino Hachado Copahue Bobadilla Calabozos Maipo Waven Wotten Trois Morne

195

------ref. 7 Caldera ------ref. 6 26511 15611 24053 94922 72245 Caldera ------ref. 5 94922 94922 94922 23212 94049 22652 Caldera - - - - - 211 3117 3117 3117 8920 7496 ref. 4 42728 12199 94832 63677 19888 10863 42155 94922 55571 83878 Caldera - - - 136 136 246 246 3117 3117 4183 9700 ref. 3 77113 15734 13887 21395 10863 23241 19753 83940 22652 39854 Caldera - 211 136 136 108 242 242 1314 ref. 2 45451 45738 43294 19753 19753 19753 58698 18430 22652 22652 22652 22652 88814 Caldera 215 108 108 108 108 108 108 108 108 108 108 108 287 ref. 1 97525 56090 50459 29891 21023 21023 30889 79342 Caldera TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE FALSE FALSE FALSE structures? Pre-existing 7 6 6 6 6 6 6 30 63 63 63 42 42 42 42 42 10 10 145 145 (Ma) duration Total arc Total 5 5 7 7 7 6 6 6 6 6 6 5 5 10 10 10 10 10 10 10 10 (Ma) duration Current arc 4 4 4 4 4 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 code crust Oldest 2 2 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 1 1 -1 -1 code Plate stress Caldera name Santorini Kos Ranau Maninjau Toba Anak Segara Bratan Batur Dano Krakatau Taal Laguna de Bay Irosin Kikai Ata Aira Kakuto Aso Shishimuta Pauzhetka Ksudach

196

------7 82126 15437 Caldera reference ------6 61960 80667 80667 15437 25980 94922 Caldera reference ------5 50421 87427 12245 25832 85372 89053 14029 22533 85085 22533 14029 15437 22533 Caldera reference - - - - - 4 211 108 246 246 242 4725 87427 25832 44037 22533 14029 15437 22533 14029 14029 Caldera reference - 3 211 246 245 242 246 246 108 108 22218 61960 62000 31852 22533 16666 22533 14029 16666 16666 16666 Caldera reference - 2 211 211 211 242 108 108 108 136 6095 6095 11867 11867 11867 11867 27403 31852 16666 15437 16666 Caldera reference 1 242 242 242 242 451 242 242 136 108 136 242 242 136 108 136 108 25832 12245 44037 31852 Caldera reference TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE structures? Pre-existing 145 145 145 145 145 145 145 145 145 100 100 100 100 100 100 100 100 100 100 100 (Ma) duration Total arc Total 5 5 5 5 5 5 5 5 5 25 25 25 25 25 25 25 25 25 25 25 (Ma) duration Current arc 3 3 3 3 3 3 3 3 3 2 2 1 1 1 1 1 1 1 1 1 code crust Oldest 1 1 1 1 1 1 1 1 1 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -1 code stress Upper plate Caldera name Opala Gorely Karymshina Akademii Karymsky Stena-Soboliny Bolshoi Sem. Uzon Krasheninnikov Akan Kutcharo Golovnin Mendeleev L'vinaya Past Urbich Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr

197

------7 28565 52414 Caldera reference ------6 35459 44601 63891 27935 27935 27935 27935 Caldera reference ------5 18117 22533 75495 96840 24053 25107 27788 27788 93981 27788 27788 78901 Caldera reference - - - 4 94 11342 11342 35459 96840 63891 96840 63891 94195 14974 14974 14974 14974 14974 13591 Caldera reference - - 3 5513 4295 4295 4295 4295 4295 11342 16666 15836 25981 96840 63891 49716 18242 63891 89349 44321 Caldera reference 2 94 136 108 279 108 5184 1306 1306 11477 43947 63891 23524 22760 22760 22760 22649 68526 29220 29220 Caldera reference 1 108 136 108 136 108 108 407 11342 11477 22470 47529 46455 16841 22649 36805 27403 46307 48347 46307 Caldera reference TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE FALSE structures? Pre-existing 4 66 66 66 66 66 66 66 66 66 66 66 66 43 43 43 43 43 100 (Ma) duration Total arc Total 4 25 13 13 13 13 13 13 13 13 13 13 13 13 43 43 43 43 43 (Ma) duration Current arc 1 4 3 3 3 3 3 3 3 3 3 3 3 1 1 1 1 1 2 code crust Oldest 2 2 2 2 2 3 -1 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -2 -2 code stress Upper plate Caldera name Nemo Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll Higashi Kurose Hole Witori-Pago

198

------7 98654 Caldera reference ------6 159 69763 34719 80895 Caldera reference ------5 24 108 4538 49938 94922 Caldera reference ------7 4 72 64 8200 1340 74300 28198 10350 10350 94922 12288 Caldera reference - - - - - 9 3 21 108 108 3820 10407 69763 94922 58433 42753 48225 15824 41880 44696 Caldera reference - 9 2 46 136 1178 8259 6681 84468 69763 38668 80822 94372 90443 44837 42753 42753 89974 74913 48205 Caldera reference 4 1 37 18 136 136 108 108 108 1178 74300 34845 80822 42753 58071 42753 21789 89524 42753 42753 Caldera reference TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE structures? Pre-existing 4 4 4 23 23 12 12 12 12 12 12 12 55 42 42 42 35 35 (Ma) duration Total arc Total 4 4 4 2 2 2 2 2 2 2 7 7 25 23 23 55 42 42 42 (Ma) duration Current arc 2 2 2 1 1 1 4 4 4 4 4 4 4 2 3 3 3 4 4 code crust Oldest 3 3 3 3 2 2 2 2 2 2 2 2 2 0 0 0 1 1 -1 code stress Upper plate Caldera name Lolobau Rabaul Tavui Kuwae Macauley Denham Taupo Whakamaru Reporoa Mangakino Kapenga Okataina Rotorua Semisopochnoi Fisher Emmons Lake Aniakchak Crater Lake Newberry

199

------7 48583 51846 Caldera reference ------6 61 27359 98993 Caldera reference ------5 98 20 141 30411 65852 69857 16555 28442 41788 69865 Caldera reference - - - - 4 98 52 94 267 267 108 7522 3501 26554 23805 46265 12199 42728 64410 22831 54794 22831 Caldera reference - - 3 94 20 20 278 186 267 186 267 135 194 186 186 5824 5824 49119 11521 61123 48209 35902 Caldera reference - 2 59 60 108 186 261 108 108 318 108 108 8259 8869 5537 5913 13923 13923 26631 72954 13779 25720 Caldera reference 1 98 58 12 263 267 273 135 136 136 212 136 244 108 5537 1498 3193 13923 35902 64410 48650 33696 Caldera reference TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE FALSE FALSE FALSE FALSE structures? Pre-existing 35 38 60 60 60 60 60 60 60 60 60 60 60 50 180 180 180 185 185 185 (Ma) duration Total arc Total 7 38 38 12 12 12 12 12 12 12 12 12 12 12 25 25 25 25 26 26 26 (Ma) duration Current arc 4 5 3 4 4 4 4 4 4 4 3 3 3 3 1 4 4 4 4 5 5 code crust Oldest 0 1 1 0 0 0 0 0 0 0 0 0 0 0 -3 -3 -3 -3 -3 -3 -3 code stress Upper plate Caldera name Kulshan Humeros, Los Primavera, La Berlin Carbonera Ilopongo Boqueron Coatepeque Amatitlan Atitlan Guayabo Guachipelin Apoyo Pelona El Valle Barrancas Chalupas Chacana Incapillo Robledo Purico

200

------7 3287 2481 Caldera reference - - - 6 873 2950 84647 47739 14249 Caldera reference - - 5 5305 1618 17516 25300 14249 21291 Caldera reference - 4 211 29580 27962 23587 22380 17273 Caldera reference - 3 3287 27962 14470 27962 50068 38514 35902 Caldera reference - 2 136 108 136 136 27962 18835 38514 Caldera reference - 1 108 15730 18271 39348 46942 21291 41341 Caldera reference TRUE TRUE TRUE TRUE FALSE FALSE FALSE FALSE structures? Pre-existing 23 23 165 165 165 165 165 165 (Ma) duration Total arc Total 15 15 15 15 15 15 23 23 (Ma) duration Current arc 4 4 4 4 4 4 1 1 code crust Oldest 1 1 -1 -1 -1 -1 -1 -1 code stress Upper plate Caldera name Cord. Nevada Pino Hachado Copahue Bobadilla Calabozos Maipo Waven Wotten Trois Morne

201

" Notes Obliquities and velocities calculated interpolated from Slab dip from 15561, obliquities calculated 78879 Slab dip from 15561, obliquities calculated 78879 Dips from 15561. Obliquities calculated 78885 model; Backarc 53643; Slab dip from 16030. Trench-normal convergence rate from HS3-NUVEL 1A plate 1A convergence rate from HS3-NUVEL Trench-normal ------324 324 26511 26511 35888 35888 35888 Tectonic Tectonic reference 4 ------238 238 5590 5590 5590 28733 28733 75292 75292 34347 34347 34347 34347 34347 34347 Tectonic Tectonic reference 3 279 279 279 279 279 279 234 234 5590 5590 1314 1314 45738 45738 30958 30958 30958 35888 35888 35888 91515 Tectonic Tectonic reference 2 324 324 9700 9700 97525 56090 19753 19753 19753 19753 19753 19753 19753 19753 30889 22652 22652 22652 22652 22652 22652 Tectonic Tectonic reference 1 ------ref. 8 Caldera name Caldera Santorini Kos Ranau Maninjau Toba Anak Segara Bratan Batur Dano Krakatau Taal Laguna de Bay Irosin Kikai Ata Aira Kakuto Aso Shishimuta Pauzhetka Ksudach

202

Notes ------324 324 324 324 324 324 324 324 324 Tectonic Tectonic reference 4 - - 238 238 238 238 238 238 238 238 238 16666 16666 16666 16666 16666 16666 16666 16666 16666 Tectonic Tectonic reference 3 ------234 234 234 234 234 234 234 234 234 16666 16666 Tectonic Tectonic reference 2 324 324 324 324 324 324 324 324 324 44037 44037 16666 16666 16666 16666 16666 16666 16666 16666 16666 Tectonic Tectonic reference 1 ------8 Caldera reference name Caldera Opala Gorely Karymshina Akademii Karymsky Stena-Soboliny Bolshoi Sem. Uzon Krasheninnikov Akan Kutcharo Golovnin Mendeleev L'vinaya Past Urbich Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr

203

Notes Slab dip from 1556, slab age 26793, obliquity 2498 ------88259 88259 88259 88259 88259 88259 32815 Tectonic Tectonic reference 4 - - - - - 16666 22760 22760 22760 22760 22760 22760 22760 16666 16666 16666 16666 16666 12894 Tectonic Tectonic reference 3 - 279 279 279 279 279 279 279 279 279 279 23241 23241 14974 14974 14974 14974 14974 69763 Tectonic Tectonic reference 2 279 279 3385 4308 4308 4308 4308 4308 11342 11342 11342 11342 11342 11342 16666 22649 22649 22649 10407 Tectonic Tectonic reference 1 ------8 72807 Caldera reference name Caldera Nemo Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll Higashi Kurose Hole Witori-Pago

204

" " " " " model; Back-arc & Slab Age = Kermadec from 15561 model; Back-arc & Slab Age = Kermadec from 15561 model; Back-arc & Slab Notes plate 1A convergence rate from HS3-NUVEL Trench-normal plate 1A convergence rate from HS3-NUVEL Trench-normal ------166 166 166 32815 32815 32815 40901 Tectonic Tectonic reference 4 ------79 79 79 72 285 12894 12894 12894 17164 17164 40901 Tectonic Tectonic reference 3 78 78 78 78 64 8259 69763 69763 69763 49938 34845 34845 67315 67315 67315 67315 67315 67315 67315 Tectonic Tectonic reference 2 46 64 285 285 285 6681 10407 10407 10407 47270 17164 17164 87667 87667 87667 87667 87667 87667 87667 Tectonic Tectonic reference 1 ------8 Caldera reference name Caldera Lolobau Rabaul Tavui Kuwae Macauley Denham Taupo Whakamaru Reporoa Mangakino Kapenga Okataina Rotorua Semisopochnoi Fisher Emmons Lake Aniakchak Crater Lake Newberry

205

" " Notes Dip from 87328, slab age 89687 velocity and obliquity calculated from HS3- Trench-normal NUVEL 1A plate model and 85459. Slab dips from 15561. 1A NUVEL ------40901 27901 51494 51494 51494 51494 51494 51494 51494 51494 51494 48154 Tectonic Tectonic reference 4 - - 72 69 5537 5537 5537 5537 5537 5537 5537 27901 14930 14930 40151 40151 40151 25224 48154 48154 Tectonic Tectonic reference 3 53 153 267 267 267 267 267 267 267 198 198 198 8259 5537 5537 11521 11521 11521 51494 51494 51494 Tectonic Tectonic reference 2 98 286 153 186 186 186 186 186 186 186 186 186 5147 5147 5147 14930 14930 51071 46783 46783 46783 Tectonic Tectonic reference 1 ------8 94818 Caldera reference name Caldera Kulshan Humeros, Los Primavera, La Berlin Carbonera Ilopongo Boqueron Coatepeque Amatitlan Atitlan Guayabo Guachipelin Apoyo Pelona El Valle Barrancas Chalupas Chacana Incapillo Robledo Purico

206

Notes ------48154 48154 Tectonic Tectonic reference 4 10071 10071 10071 10071 10071 10071 25098 25098 Tectonic Tectonic reference 3 167 167 167 167 21022 21022 17486 17486 Tectonic Tectonic reference 2 21022 21022 21022 21022 27962 27962 41341 41341 Tectonic Tectonic reference 1 ------8 Caldera reference name Caldera Cord. Nevada Pino Hachado Copahue Bobadilla Calabozos Maipo Waven Wotten Trois Morne

207

8 Appendix B

Regional and caldera-specific parameters for calderas examined in Chapter 3.

Codes • Tectonic setting: 1 = arc, 2 = ocean hotspot, 3 = intracontinental • Crust type: 1 = oceanic, 2 = transitional, 3 = continental • Caldera-forming eruption composition: 0 = trachyte, 1 = dacite, 2 = rhyodacite, 3 = rhyolite, 4 = peralkaline rhyolite • Local stress code: -2 = compression, -1 = transpression, 0 = neutral 1 = transtension, 2 = extension

Notes • “Date of formation” refers to first CFE when known. • Numbers in the reference column contain index numbers that correspond to the citations listed in Appendix C. • Additional reference information for calderas in arcs can be found in Appendix A. • “Classification” columns show the classification for the three regional classes output by the naïve Bayes classifier. Bold numbers indicate that the output matches the true value. Note that in classification of stress regime, 1 = neutral, compression or transpression and 2 = transtension or extension. • “Parameters available” refers to the number of the parameters listed in Table 5 that are available for the caldera.

208

0 9.1 3.4 211 238 510 200 500 980 450 132 950 300 1300 1600 2000 (ky) Date of Formation 8 6 8 9 15 20 18 12 16 20 9.5 5.5 9.5 14.5 14.5 26.5 Average Diameter 1 1 3 2 2 1 3 3 3 3 3 1 3 3 2 2 CFE Composition 1 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 type Crust -68.830 -67.740 -67.750 -78.322 -78.390 -78.436 -69.780 -72.220 -71.170 -71.060 -70.507 -70.496 179.607 -164.386 -162.080 -158.145 Longitude -0.787 -0.270 -0.677 51.944 54.668 55.340 56.904 -27.900 -26.760 -23.000 -34.171 -40.500 -38.670 -37.837 -36.063 -35.558 Latitude 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location - US Alaskan Islands (Aleutians) Alaskan Peninsula - US Alaskan Peninsula - US Alaskan Peninsula - US Argentina Andes C. - Argentina Andes C. - Andes C. - Chile Andes N. - Ecuador Andes N. - Ecuador Andes N. Ecuador Argentina Andes S. - Andes S. - Chile Andes S. - Chile Andes S. - Chile Andes S. - Chile Andes S. - Chile Caldera name Semisopochnoi Fisher Emmons Lake Aniakchak Incapillo Robledo (Cerro Blanco) Purico Complex Barrancas (Cotopaxi I) Chacana Chalupas Maipo-Diamante Cordillera Nevada Pino Hachado Copahue-Caviahue Bobadilla Calabozos

209

15 12 22 84 40 72 23 6.8 500 191 100 550 665 1150 2000 1600 Date of Formation (first), (ky) 9 7 5 15 5.5 6.0 5.0 6.3 5.5 9.5 6.5 18.5 5.25 18.5 17.5 17.5 Average Diameter 0 0 0 2 2 2 3 3 1 1 2 1 2 3 3 1 CFE Composition 1 1 1 3 3 3 2 2 2 2 2 2 2 2 2 2 type Crust -25.47 -25.32 -25.78 -90.590 -91.193 -88.518 -88.930 -89.053 -89.294 -89.550 -85.153 -85.330 -86.030 -122.120 -121.229 -121.710 Longitude 37.77 37.77 37.87 11.920 42.930 43.722 48.830 14.449 14.678 13.495 13.630 13.672 13.734 13.870 10.748 10.830 Latitude 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location Nicaragua Salvador Salvador Salvador Salvador Salvador Azores Azores Azores Cascades - US Cascades - US Cascades - US America N. - Central America N. - Central America N. El Central America N. El Central America N. El Central America N. El Central America N. El Central America S. - Costa Central America S. - Costa Central America S. - Central Rica Rica Guatemala Guatemala Caldera name Salvador) Agua de Pau Furnas (Povcacau) Sete Cidades Crater Lake Newberry Kulshan Amatitlan Atitlan Berlin Carbonera Caldera, La Ilopango Boqueron (San Coatepeque Guayabo Guachipelin Apoyo

210

0 0 7 0 0 0 29 21 240 276 224 400 0.96 2000 2000 1850 Date of Formation (first), (ky) 6 8 10 5.0 8.0 8.0 5.0 6.0 6.5 8.0 13.5 14.0 35.0 16.0 20.0 10.0 Average Diameter 1 3 3 4 4 4 4 4 4 4 4 4 4 0 0 0 CFE Composition 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 type Crust 38.6 40.2 36.1 17.17 16.42 38.58 128.08 38.397 39.174 41.602 41.658 41.699 36.353 -86.935 130.820 147.000 Longitude 7.9 7.47 -0.23 20.97 21.05 41.98 7.186 8.358 13.02 -1.157 12.677 13.095 32.030 13.284 13.371 44.600 Latitude 1 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 setting Tectonic Country/Location Nicaragua Central America S. - Central Chad Chad China/Korea African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Ethiopia East African Rift - Kenya rift East African Rift - Kenya rift East African Rift - Kenya rift East Afar Afar Afar alalta (Pruvost) ! Caldera name followed by Tousside followed by CCDB) Pelona Voon Tarso caldera Yirrigue (Baitoushan) Tianchi (Corbetti) Awasa O'a Caldera (Shalla in Gademotta Gedemsa Caldera Asavyo K'one Mallahle (Sorkale, Bidu) Mal Nabro Suswa Menengai Longonot (Gadamsa)

211

0 0 12 80 10 52 55 55 0.8 1.6 172 161 100 550 840 29.3 Date of Formation (first), (ky) 7 9 6 12 15 18 14 45 8.5 9.0 9.0 6.0 8.5 12.0 15.5 7.25 Average Diameter 2 3 3 4 3 3 1 1 1 2 1 3 3 3 4 0 CFE Composition 3 3 1 1 1 1 2 2 2 2 2 3 3 3 3 3 type Crust 13.9 12.02 -19.15 -16.75 -16.78 25.396 27.093 98.830 116.414 115.130 115.380 130.308 106.000 105.423 103.920 100.180 Longitude 63.93 65.03 65.73 2.580 36.77 40.73 -8.408 -8.280 -8.240 -6.300 -6.102 -4.830 -0.330 36.404 36.676 30.789 Latitude 1 1 2 2 2 2 1 1 1 1 1 1 1 1 3 3 setting Tectonic Country/Location Greece Greece Iceland Iceland Iceland Iceland Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Indonesia - Java Bali Indonesia - Sumatra Indonesia - Sumatra Indonesia - Sumatra Italy - Sicily margin Tyrrhenian Italy - Caldera name Santorini Kos Katla Torfajokull Askja Krafla Anak Segara Bratan Batur Dano (Danau or Krakatau Ranau Maninjau Toba Pantelleria (Cinque Ischia Denti) Banten)

212

0 0 36 73 55 41 30 1.7 760 130 600 160 1300 1500 2000 2000 Date of Formation (first), (ky) 6 8 8 6 11 18 13 10 14 13 25 6.5 8.5 6.5 11.5 13.5 Average Diameter 0 2 3 1 2 2 2 2 3 3 3 2 2 2 1 3 CFE Composition 3 3 2 2 2 2 2 2 2 2 2 2 2 1 1 1 type Crust 14.14 139.845 140.730 140.518 140.695 140.881 140.770 140.880 140.854 141.320 140.820 143.150 142.880 140.050 140.010 139.850 Longitude 40.83 37.194 38.730 38.736 38.830 40.469 40.579 40.650 42.599 42.750 43.080 43.516 43.680 31.470 31.950 32.100 Latitude 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location Italy - Tyrrhenian margin Tyrrhenian Italy - Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan - Northeast Honshu Japan-Izu Islands Japan-Izu Islands Japan-Izu Islands Caldera name Campi Flegrei Hatori Narugo Akakura Mukaimachi - Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi-Mitsumata Daisetsu Sumisu Myojinsho Myojin Knoll

213

0 0 85 22 77 39 50 40 40 110 300 443 180 450 560 1000 1780 Date of Formation (first), (ky) 6 8 9 3 20 20 21 19 13 15 18 10 7.5 12.5 14.0 22.5 11.25 Average Diameter 2 1 3 3 3 1 1 2 3 3 2 3 1 1 3 1 3 CFE Composition 1 1 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 type Crust 39.69 157.53 131.110 139.920 139.660 176.272 130.679 130.710 131.200 156.930 157.530 157.332 158.030 159.464 159.452 159.593 160.020 Longitude 8.8 54.85 54.115 32.450 33.400 31.345 31.650 32.880 33.167 51.430 51.806 52.513 52.558 53.981 54.071 54.320 -38.252 Latitude 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location Japan-Izu Islands Japan-Izu Islands Japan-Kyushu Japan-Kyushu Japan-Kyushu Japan-Kyushu Japan-Kyushu Japan-Kyushu -Ryukyu Kamchatka - Sredinny arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic Islands Range Caldera name Semiachik Higashi-aogashima Kurose Hole Kakuto Ata Aira Aso Shishimuta Kikai Hangar (Khangar) Pauzhetka- Kurile Lake Ksudach Massif Opala Gorely Akademii--Polovinka Karymshina Stena-Soboliny Maly- Bolshoi Semiachik

214

0 0 0 0 0 40 39 39 6.3 2.2 9.4 7.5 340 400 39.5 31.5 1260 Date of Formation (first), (ky) 5 6 8 7 6 8 10 12 19 23 16 10 10 6.5 6.5 8.5 10.5 Average Diameter 3 1 2 2 1 3 2 2 2 1 1 1 3 2 2 1 2 CFE Composition 3 3 3 1 1 2 2 1 1 1 1 1 1 1 1 1 1 type Crust -19.03 159.970 160.273 144.013 144.430 121.270 145.501 145.740 147.200 148.220 148.800 150.880 151.950 154.710 154.810 -178.431 -177.964 Longitude 63.63 54.500 54.593 43.384 43.550 14.400 43.865 43.980 44.625 45.192 45.380 46.520 46.920 49.360 49.570 -30.232 -29.270 Latitude 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic arc Kamchatka - Volcanic Kermadec Islands Kermadec Islands Kurile - N. Japan Kurile - N. Japan Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Kurile Islands Caldera name Uzon-Geyzernaya Krasheninnikov Karymsky Macauley Island Denham Akan Kutcharo (Mashu) L'vinaya Past Golovnin Mendeleev Urbich Isthmus Vetrovoi Medvezhii or Medvezhia Chirpoi Zavaritsky Tao-Rusyr Nemo

215

30 40 95 27 12 1.6 3.3 1.4 7.1 460 340 240 280 220 900 0.89 1300 Date of Formation (first), (ky) 6 8 11 11 10 18 22 35 12 22 43 21 20 15 5.3 6.5 10.5 Average Diameter 1 1 3 4 3 3 3 3 3 3 3 1 1 2 3 2 2 CFE Composition 1 1 3 3 3 3 3 3 3 3 3 1 2 2 2 2 2 type Crust -80.17 -118.87 -61.317 -61.330 -97.450 175.925 176.061 176.330 175.782 176.501 176.244 157.968 150.520 152.203 152.200 151.158 -103.500 Longitude 8.58 37.7 -5.581 -4.271 -4.122 -4.920 15.330 15.370 19.680 20.670 52.780 -38.815 -38.553 -38.420 -38.364 -38.182 -38.085 Latitude 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 setting Tectonic Country/Location Lesser Antilles Lesser Antilles Lesser Mexico Mexico New Zealand New Zealand New Zealand New Zealand New Zealand New Zealand New Zealand Panama Papua New Guinea Papua New Guinea Papua New Guinea Papua New Guinea Philippines N. Caldera name Wotten Waven Wotten Pitons Trois Morne Humeros, Los Primavera, La Taupo Whakamaru-Maroa Reporoa Mangakino Okataina Caldera Rotorua Kapenga El Valle Witori-Pago Rabaul Tavui Lolobau Laguna de Bay Complex

216

37 35 0.5 140 100 272 760 1450 1300 Date of Formation (first), (ky) 9 11 18 5.0 5.5 7.0 24.5 22.0 65.0 Average Diameter 1 3 0 3 4 3 3 3 1 CFE Composition 2 2 3 3 3 3 3 3 1 type Crust 24.27 34.52 42.23 36.431 -110.67 -106.57 121.000 124.050 168.540 Longitude 12.95 38.57 38.65 35.87 44.43 -0.926 14.000 12.770 -16.830 Latitude 1 1 3 3 3 3 3 3 1 setting Tectonic Country/Location Philippines N. Philippines S. Sudan Turkey Turkey US-California US-New Mexico US-Wyoming (New Hebrides) Vanuatu Caldera name Taal Irosin Deriba Acigol Nemrut Dagi Long Valley Caldera Valles Yellowstone Kuwae

217

- - - - 2.51 2.22 3.47 2.97 3.80 3.89 3.30 3.12 4.61 5.67 4.38 5.18 K2O (wt%) - - - - 4.60 4.47 3.74 3.67 4.45 4.05 5.17 4.82 5.68 5.58 4.08 3.68 Na2O (wt%) - - - - 0.71 0.74 0.34 0.52 0.37 0.11 0.12 0.10 0.20 0.47 0.38 0.48 TiO2 (wt%) - - - - 71.60 70.27 71.51 68.24 68.91 75.82 70.41 62.13 65.76 76.28 74.48 73.73 SiO2 (wt%) 61.03 67.89 75.00 69.90 71.51 65.20 66.40 74.00 75.00 73.00 74.29 63.00 71.85 74.10 71.00 70.50 High SiO2 CFE (wt%) 27 33 37 33 70 70 70 50 50 50 55 32 32 32 32 47 Crustal thickness 1 1 1 1 2 2 2 2 2 2 1 2 2 2 2 2 regime Local stress 2 1 1 2 2 2 1 1 1 2 2 -2 -2 -2 -2 -2 regime Local stress 0 0 0 7 4 10 1.6 5.6 1.7 2.7 1.7 4.9 4.3 0.01 0.595 0.377 (Ma) Local volcanism duration of Caldera name Semisopochnoi Fisher Emmons Lake Aniakchak Incapillo Robledo Purico Complex Barrancas Chacana Chalupas Maipo Cordillera Nev. Pino Hachado Copahue Bobadilla Calabozos

218

- 4.54 5.52 2.80 2.20 2.21 1.39 2.11 4.37 3.82 4.42 2.19 5.73 3.96 2.28 2.36 K2O (wt%) - 6.25 4.84 5.61 4.35 2.87 5.19 4.31 3.12 2.67 2.49 4.52 6.56 7.43 3.78 2.96 Na2O (wt%) - 0.25 0.29 0.44 0.44 0.31 0.17 0.67 0.51 0.11 0.21 0.25 0.50 0.76 0.36 0.33 TiO2 (wt%) - 69.61 63.31 63.19 73.17 69.27 73.61 73.89 64.90 65.71 69.39 72.69 77.29 73.60 67.42 75.83 SiO2 (wt%) - 69.00 66.00 62.92 72.00 72.40 73.30 74.70 75.84 65.36 65.72 69.20 69.90 77.00 76.00 66.00 High SiO2 CFE (wt%) 14 14 14 40 37 40 47 47 35 36 36 37 37 37 37 33 Crustal thickness 2 2 2 2 2 1 2 2 2 2 2 2 1 2 2 2 regime Local stress 2 2 2 2 2 1 1 2 2 1 1 2 2 1 -2 -1 regime Local stress 0 0 2 0 8 8 14 0.4 0.5 0.8 1.29 0.59 0.07 0.59 0.181 0.093 (Ma) Local volcanism duration of Caldera name Agua de Pau Furnas Sete Cidades Crater Lake Newberry Kulshan Amatitlan Atitlan Berlin Carbonera Ilopango Boqueron Coatepeque Guayabo Guachipelin Apoyo

219

------4.30 4.82 4.94 4.61 4.33 4.63 4.48 4.76 5.23 K2O (wt%) ------4.60 5.47 5.29 6.82 5.77 5.49 7.92 6.58 7.48 Na2O (wt%) ------0.22 0.37 0.41 0.37 0.15 0.51 0.62 0.62 0.33 TiO2 (wt%) ------73.47 68.84 68.02 73.02 63.02 76.28 75.53 62.06 66.88 SiO2 (wt%) ------75.74 75.53 73.47 68.84 64.86 72.08 61.03 68.00 62.00 High SiO2 CFE (wt%) 35 27 27 40 28 28 28 28 25 28 28 35 25 35 35 35 Crustal thickness 2 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 regime Local stress 2 0 0 1 1 2 2 2 1 2 2 2 2 2 2 2 regime Local stress 0 0 0 3 0 0 0 0.4 0.4 3.69 0.21 0.21 0.21 0.21 0.55 0.18 (Ma) Local volcanism duration of alalta ! Caldera name Pelona Voon Tarso Yirrigue Tianchi Awasa O'a Caldera Gademotta Gedemsa Asavyo K'one Mallahle Mal Nabro Suswa Menengai Longonot

220

- - 3.60 4.10 3.50 3.91 2.57 2.51 3.50 2.01 4.52 5.09 4.44 6.50 4.26 2.36 K2O (wt%) - - 5.32 3.69 5.45 5.50 4.40 3.65 5.12 3.74 3.57 6.04 7.74 4.96 4.58 3.36 Na2O (wt%) - - 0.42 0.17 0.31 0.27 0.29 0.61 0.54 0.75 0.25 0.12 0.60 0.51 0.56 0.18 TiO2 (wt%) - - 70.09 75.59 70.97 68.55 70.42 73.45 74.39 68.17 61.67 73.36 75.58 72.68 66.23 77.16 SiO2 (wt%) - - 72.00 77.00 71.60 74.00 75.00 73.00 66.22 68.00 64.00 73.00 74.00 74.39 67.98 62.00 High SiO2 CFE (wt%) 32 27 20 22 32 20 20 20 20 20 20 30 30 30 18 22 Crustal thickness 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 2 regime Local stress 2 2 2 2 2 2 2 2 2 1 1 1 1 2 1 -1 regime Local stress 0 0 0 0 0 0 0 0 0 1.6 3.4 0.5 1.3 0.15 0.384 0.012 (Ma) Local volcanism duration of Caldera name Santorini Kos Katla Torfajokull Askja Krafla Anak Segara Bratan Batur Dano Krakatau Ranau Maninjau Toba Pantelleria Ischia

221

- - 8.97 1.54 1.32 1.17 1.81 1.27 2.55 3.62 3.51 0.90 0.79 0.81 1.76 3.58 K2O (wt%) - - 4.44 3.25 4.30 2.85 4.41 4.17 5.04 4.01 4.49 3.88 3.43 3.76 4.03 4.48 Na2O (wt%) - - 0.42 0.39 0.45 0.51 0.29 0.49 0.32 0.11 0.30 0.47 0.62 0.32 0.28 0.08 TiO2 (wt%) - - 72.64 75.01 68.50 73.19 69.91 76.41 74.17 78.07 66.77 75.02 61.26 74.63 74.08 72.53 SiO2 (wt%) 63.00 72.00 75.01 64.00 72.00 72.00 72.18 72.00 76.90 74.00 77.31 71.57 72.00 71.12 66.40 75.02 High SiO2 CFE (wt%) 24 30 30 30 30 30 30 30 30 30 30 34 40 23 23 23 Crustal thickness 2 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 regime Local stress 1 2 2 2 2 2 -2 -2 -2 -2 -2 -2 -2 -1 -2 -2 regime Local stress 0 0 0 0 0 0 0 0 0 0 0 4.8 0.4 3.5 2.8 0.06 (Ma) Local volcanism duration of Caldera name Campi Flegrei Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll

222

- - - 1.32 1.32 2.14 3.49 1.65 1.57 1.54 2.62 2.31 2.79 2.93 3.83 2.98 2.68 K2O (wt%) - - - 4.90 3.50 3.89 4.12 4.44 4.25 4.91 4.27 4.09 4.81 4.05 4.48 3.66 5.36 Na2O (wt%) - - - 0.52 0.39 0.17 0.90 0.70 0.55 0.25 0.94 0.30 0.74 0.21 0.26 0.28 0.58 TiO2 (wt%) - - - 76.59 68.31 64.39 73.64 71.77 74.54 65.64 73.41 68.07 75.84 74.98 69.98 73.18 67.86 SiO2 (wt%) 73.17 68.00 75.00 74.00 76.50 66.00 64.90 72.00 72.85 72.31 71.78 76.00 65.00 65.41 71.86 67.20 74.44 High SiO2 CFE (wt%) 23 23 30 30 30 30 30 30 30 30 32 37 32 37 32 37 37 Crustal thickness 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 2 2 regime Local stress 2 2 2 2 2 2 2 2 1 1 2 1 1 1 1 1 -1 regime Local stress 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0.7 4.2 0.93 Local volcanism duration of Caldera name Higashi Kurose Hole Kakuto Ata Aira Aso Shishimuta Kikai Hangar Pauzhetka Ksudach Opala Gorely Akademii Karymshina Stena-Soboliny Bolshoi Semi.

223

- - - 2.32 0.67 2.02 1.67 0.65 1.12 2.19 0.81 4.19 2.83 1.63 1.78 1.36 1.53 K2O (wt%) - - - 4.39 4.49 3.34 4.20 3.70 4.01 3.41 3.51 3.97 4.59 4.61 4.20 4.46 4.16 Na2O (wt%) - - - 0.40 0.51 0.65 0.55 0.61 0.84 0.55 0.60 0.70 0.69 0.65 0.63 0.76 0.63 TiO2 (wt%) - - - 74.02 68.09 70.71 73.35 74.07 66.44 69.32 65.39 68.59 68.38 72.23 69.83 69.43 72.88 SiO2 (wt%) - - 73.34 69.58 72.00 67.70 72.43 71.92 70.00 67.00 67.00 67.81 74.00 72.00 72.00 67.00 64.00 High SiO2 CFE (wt%) 37 37 37 15 15 28 28 20 25 25 20 21 21 17 17 25 25 Crustal thickness 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 regime Local stress 1 1 1 2 2 2 1 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 regime Local stress 0 0 0 0 0 0 0 0 0 0 0 0 2.5 1.4 3.7 1.6 1.03 Local volcanism duration of Caldera name Uzon Krasheninnikov Karymsky Macauley Denham Akan Kutcharo L'vinaya Past Golovnin Mendeleev Urbich Isthmus Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr Nemo

224

- 1.59 5.17 4.35 2.41 4.20 2.72 1.17 3.91 1.71 1.29 5.05 4.86 2.83 3.78 5.88 1.53 K2O (wt%) - 3.64 4.84 4.01 3.21 4.60 3.71 4.57 4.29 4.35 4.17 4.10 3.26 3.76 3.08 3.83 4.56 Na2O (wt%) - 0.42 0.05 0.09 0.21 0.14 0.24 0.29 0.39 0.80 0.32 0.74 0.33 0.16 0.33 0.28 0.53 TiO2 (wt%) - 76.07 75.72 75.99 74.27 76.32 68.11 68.41 67.17 69.52 67.24 63.58 76.98 75.68 71.28 73.58 75.28 SiO2 (wt%) - 62.40 77.00 77.00 76.00 77.00 75.00 76.00 76.00 76.00 73.00 68.70 66.26 64.82 75.00 69.20 69.00 High SiO2 CFE (wt%) 30 30 42 26 20 20 20 20 20 20 20 22 30 32 32 30 34 Crustal thickness 2 2 2 2 2 2 2 2 2 2 2 1 1 2 2 1 2 regime Local stress 2 2 2 2 1 1 2 1 1 1 1 1 1 2 -2 -2 -2 regime Local stress 0 0 0 0 0 2 3.5 0.3 1.7 1.7 1.7 1.7 1.7 1.7 0.12 1.55 0.19 Local volcanism duration of Caldera name Pitons Wotten Waven Wotten Trois Morne Humeros, Los Primavera, La Taupo Whakamaru Reporoa Mangakino Okataina Rotorua Kapenga El Valle Witori-Pago Rabaul Tavui Lolobau Laguna de Bay

225

3.07 4.05 4.42 4.74 5.15 4.72 5.59 2.62 3.03 K2O (wt%) 4.00 4.42 5.55 3.35 4.51 2.87 6.23 4.16 3.93 Na2O (wt%) 0.17 0.09 0.25 0.09 0.17 0.73 0.88 0.13 0.58 TiO2 (wt%) 64.61 75.52 72.89 77.91 74.90 64.51 76.06 63.13 77.66 SiO2 (wt%) 64.29 75.60 65.00 75.52 74.16 77.00 77.64 77.00 73.00 High SiO2 CFE (wt%) 34 45 39 42 45 42 37 42 28 Crustal thickness 2 2 2 1 2 2 2 2 2 regime Local stress 1 1 1 1 1 2 2 1 -1 regime Local stress 0 0 0 10 10 1.1 4.5 2.1 14.5 Local volcanism duration of Caldera name Taal Irosin Deriba Acigol Nemrut Dagi Long Valley Caldera Valles Yellowstone Kuwae

226

Caldera name Semisopochnoi Fisher Emmons Lake Aniakchak Incapillo Robledo Purico Complex Barrancas Chacana Chalupas Maipo Cordillera Nev. Pino Hachado Copahue Bobadilla Calabozos ------0.704 0.7033 0.7034 0.7066 0.7042 0.7038 0.706693 0.709008 87/86 Sr - - - - 10 800 878 849 743 335 752 961 786 670 880 884.21 Ba (ppm) - - - - 69 12 92 328 255 223 468 301 362 229 141 93.9 Sr (ppm) - - - - - 56 93 146 121 173 238 196 64.1 66.2 74.2 82.69 Rb (ppm) ------9 29 13 15 79 5.1 7.3 7.82 15.18 Nb (ppm) ------8 46 10 23 17 41 15 34.3 49.8 44.03 Y (ppm) - - - - - 84 91 211 190 263 241 122 315 137 10.5 194.1 Zr( ppm) Caldera name Semisopochnoi Fisher Emmons Lake Aniakchak Incapillo Robledo Purico Complex Barrancas Chacana Chalupas Maipo Cordillera Nev. Pino Hachado Copahue Bobadilla Calabozos

227

Caldera name Agua de Pau Furnas Sete Cidades Crater Lake Newberry Kulshan Amatitlan Atitlan Berlin Carbonera Ilopango Boqueron Coatepeque Guayabo Guachipelin Apoyo ------0.7038 0.70349 0.70405 0.703867 87/86 Sr - - - - - 810 900 810 694 848 1135 1000 1323 1783 1776 1269 Ba (ppm) - - - - 67 305 180 295 160 335 363 386 318 215 257 400 Sr (ppm) - - - - 57 92 44 33 55 73 50 118 116 180 103 44.3 Rb (ppm) ------5 9 3.38 7.58 Nb (ppm) ------3 6 25 90 20 13 18 14 37.4 12.99 Y (ppm) - - - - 118 115 252 300 120 232 107 169 216 145 140 153.6 Zr( ppm) Caldera name Agua de Pau Furnas Sete Cidades Crater Lake Newberry Kulshan Amatitlan Atitlan Berlin Carbonera Ilopango Boqueron Coatepeque Guayabo Guachipelin Apoyo

228

alalta ! Caldera name Pelona Voon Tarso Yirrigue Tianchi Awasa O'a Caldera Gademotta Gedemsa Asavyo K'one Mallahle Mal Nabro Suswa Menengai Longonot ------0.708 0.7053 0.70589 0.70577 87/86 Sr ------6 66 17 10 602 Ba (ppm) ------9 3 5 45 11.7 32.2 Sr (ppm) ------184 216 167 111.8 128.6 Rb (ppm) ------285 369 246 113.9 126.7 Nb (ppm) ------91 145 210 129 62.7 Y (ppm) ------579 1174 2038 1356 1032 668.8 Zr( ppm) alalta ! Caldera name Pelona Voon Tarso Yirrigue Tianchi Awasa O'a Caldera Gademotta Gedemsa Asavyo K'one Mallahle Mal Nabro Suswa Menengai Longonot

229

Caldera name Santorini Kos Katla Torfajokull Askja Krafla Anak Segara Bratan Batur Dano Krakatau Ranau Maninjau Toba Pantelleria Ischia - - - - - 0.714 0.7032 0.7032 0.70333 0.70318 0.70394 0.70407 0.70688 0.705025 0.704461 0.703479 87/86 Sr - - - - - 463 670 501 488 544 392 326 557 208 4.39 1007 Ba (ppm) - - - - 80 46 26 111 114 149 108 216 191 184 254 5.18 Sr (ppm) - - - - 79 95 49 88 61 113 110 101 107 182 128 458 Rb (ppm) - - - - - 9 15 10 15 19 2.2 118 162 246 103 12.85 Nb (ppm) - - - - 7 11 97 40 47 56 24 96 123 105 50.7 69.5 Y (ppm) - - - - 87 35 112 909 890 437 310 266 268 815 1292 299.7 Zr( ppm) Caldera name Santorini Kos Katla Torfajokull Askja Krafla Anak Segara Bratan Batur Dano Krakatau Ranau Maninjau Toba Pantelleria Ischia

230

Caldera name Campi Flegrei Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll ------0.704 0.7075 0.7042 0.703962 0.703962 0.703682 87/86 Sr ------121 349 385 344 504 195 186 1057 224.7 Ba (ppm) ------276 223 173 202 186 127 73.7 143.3 Sr (ppm) ------9 24 35 30 25 373 13.2 Rb (ppm) ------7 4 3 1 3 43 6.7 Nb (ppm) ------41 51 37 34 30.9 49.3 33.09 Y (ppm) ------333 142 161 125 127 126 75.5 Zr( ppm) Caldera name Campi Flegrei Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll

231

Caldera name Higashi Kurose Hole Kakuto Ata Aira Aso Shishimuta Kikai Hangar Pauzhetka Ksudach Opala Gorely Akademii Karymshina Stena-Soboliny Bolshoi Semi. ------0.7048 0.70592 0.70337 0.70332 0.704334 0.703255 0.703443 87/86 Sr ------372 497 359 783 801 545 741 1070 Ba (ppm) ------221 351 191 241 348 274 273 177 57.2 Sr (ppm) ------20 59 27 22 47 51 40 52 132 91.8 Rb (ppm) ------0.8 7.9 2.9 3.3 1.6 16.3 3.69 Nb (ppm) ------31 30 10 48 13 36 10 1.7 2.83 37.5 8.15 Y (ppm) ------98 87 155 282 157 103 270 183 57.2 Zr( ppm) Caldera name Higashi Kurose Hole Kakuto Ata Aira Aso Shishimuta Kikai Hangar Pauzhetka Ksudach Opala Gorely Akademii Karymshina Stena-Soboliny Bolshoi Semi.

232

Caldera name Uzon Krasheninnikov Karymsky Macauley Denham Akan Kutcharo L'vinaya Past Golovnin Mendeleev Urbich Isthmus Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr Nemo ------0.70372 0.70346 0.70407 87/86 Sr ------660 520 423 530 340 205.6 361.97 Ba (ppm) ------230 241 161 290 162.7 193.24 Sr (ppm) ------35 10 23 34.4 29.3 48.21 Rb (ppm) ------6.5 1.4 1.9 1.3 3.72 0.43 Nb (ppm) ------42 49 40.3 4.34 35.6 28.61 Y (ppm) ------116 210 184 153 79.4 4.24 Zr( ppm) Caldera name Uzon Krasheninnikov Karymsky Macauley Denham Akan Kutcharo L'vinaya Past Golovnin Mendeleev Urbich Isthmus Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr Nemo

233

Caldera name Pitons Wotten Waven Wotten Trois Morne Humeros, Los Primavera, La Taupo Whakamaru Reporoa Mangakino Okataina Rotorua Kapenga El Valle Witori-Pago Rabaul Tavui Lolobau Laguna de Bay - - - - - 0.7044 0.7058 0.7054 0.7058 0.7036 0.7035 0.70425 0.70551 0.70552 0.70529 0.70426 0.703536 87/86 Sr - - - 50 690 758 792 776 778 805 324 760 210 481 335 320 1030 Ba (ppm) - - 26 10 69 58 811 123 129 103 137 197 310 294 174 255 259 Sr (ppm) - - 78 81 24 14 64 21 21 115 145 266 101 241 145 169 149 Rb (ppm) - - - - - 7 5 9 3 1 15 12 10 49 6.9 108 18.8 Nb (ppm) - - 2 33 26 33 52 34 43 49 19 37 32 33 135 22.8 55.1 Y (ppm) - - 71 48 111 114 311 137 596 150 196 244 177 268 192 139 400 Zr( ppm) Caldera name Pitons Wotten Waven Wotten Trois Morne Humeros, Los Primavera, La Taupo Whakamaru Reporoa Mangakino Okataina Rotorua Kapenga El Valle Witori-Pago Rabaul Tavui Lolobau Laguna de Bay

234

Caldera name Calderas used to test geochem (not included in main database) References are given in the main text, Table 6 given in the main text, Table are References Taal Irosin Deriba Acigol Nemrut Dagi Long Valley Caldera Valles Yellowstone Kuwae - - 0.705 0.7039 0.7092 0.7109 0.70472 0.70897 0.705852 87/86 Sr - - - 793 628 316 765 217 490 Ba (ppm) - - 6 60 43 43 175 352 32.5 Sr (ppm) - - 88 79 114 117 178 180 178 Rb (ppm) - - - 7 77 16 53 44 48.03 Nb (ppm) - - - 15 41 20 37 55 70.03 Y (ppm) - - 110 119 100 492 750 215 217 Zr( ppm) Caldera name Taal Irosin Deriba Acigol Nemrut Dagi Long Valley Caldera Valles Yellowstone Kuwae

235

7 2 2 2 2 11 10 11 11 11 11 11 11 11 11 10 Available # Parameters 1 1 1 1 2 2 2 1 2 2 2 2 2 1 1 2 extensional) Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 1 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Crust Classification: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Classification: Tectonics setting Tectonics References 5240; 85901; 2202 4 37 80895 54794 61123; 141 33696 35902; 64410; 36424 51846; 27713 64410; 72954 21291 3287; 5305; 84647 2950; 2481 3287 1618 22380 Caldera name Semisopochnoi Fisher Emmons Lake Aniakchak Incapillo Robledo Purico Complex Barrancas Chacana Chalupas Maipo Cordillera Nev. Pino Hachado Copahue Bobadilla Calabozos

236

5 5 5 2 9 10 10 10 10 10 11 11 10 11 11 10 Available # Parameters 2 2 2 1 1 1 2 1 1 1 1 1 2 1 2 1 extensional) Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 1 1 1 3 3 3 3 2 2 2 3 3 3 3 3 3 Crust Classification: 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 Classification: Tectonics setting Tectonics References 136; 94922; 30573; 40768; 99018 136; 94922; 30573; 40768; 99018 136; 94922; 30573; 40768; 99018; 98654; 8259; 91273 18; 27070 48583; 94818 46265; 30411 27359; 34521 69857 26554; 267 13923; 16555 23805; 13923 12199 3501 42728 26074 Caldera name Agua de Pau Furnas Sete Cidades Crater Lake Newberry Kulshan Amatitlan Atitlan Berlin Carbonera Ilopango Boqueron Coatepeque Guayabo Guachipelin Apoyo

237

2 2 2 2 2 2 2 5 5 6 6 11 11 11 11 11 Available # Parameters 2 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 extensional) Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 2 3 3 3 3 3 3 3 3 2 3 3 3 3 3 1 Crust Classification: 1 3 3 3 3 3 3 3 3 1 3 3 3 3 1 3 Classification: Tectonics setting Tectonics 30748; 80210; 168; 8318 References 108; 94922 63192; 168; 56763; 108; 94922; 63192; 136; 70562; 108; 136; 100041; 32674; 89324; 136; 108, 94922; 29100; 41294; 136; 94922; 22203; 68362; 94225 108; 94922; 71846; 7722; 35427; 136; 94922; 68362; 94225 108; 136; 45082; 83109; 94225 108 136; 83109; 94225; 6225 136; 6225; 83109; 83513; 94225 108, 136, 6225; 83109; 83513, 94225 108; 96977; 96829; 58200 108;80193; 9880; 10266; 58200 3589; 62805 68362; 94225 68362; 94225 30748; 80210 alalta ! Caldera name Pelona Voon Tarso Yirrigue Tianchi Awasa O'a Caldera Gademotta Gedemsa Asavyo K'one Mallahle Mal Nabro Suswa Menengai Longonot

238

5 2 2 5 11 11 11 11 10 10 11 11 11 11 11 11 Available # Parameters 2 1 2 2 1 1 1 1 1 1 1 2 2 2 2 2 extensional) Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 3 3 2 1 1 1 2 2 2 2 2 3 3 3 3 3 Crust Classification: 1 1 1 2 2 2 1 1 1 1 1 1 1 1 3 3 Classification: Tectonics setting Tectonics 20700; 65957; 13524 References 91608; 83033 47561 15611 108; 15714; 88703; 7518 108; 136; 47380; 7518; 9941 136; 108; 93712; 80261; 7518 63677; 24481 50459; 61575 58698; 35291 215 43294 94832 136, 94922; 42393; 88869; 76581; 108; 94922; 136; 39560; 40041; 90083 Caldera name Santorini Kos Katla Torfajokull Askja Krafla Anak Segara Bratan Batur Dano Krakatau Ranau Maninjau Toba Pantelleria Ischia

239

5 5 2 5 5 5 9 2 9 6 11 11 11 11 11 11 Available # Parameters 2 1 1 2 1 1 1 1 1 1 1 1 2 1 1 1 extensional) Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 3 3 2 2 2 2 2 2 2 2 2 2 2 1 1 1 Crust Classification: 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Classification: Tectonics setting Tectonics References 136; 108, 94922, 4004; 10643; 6215; 75495 44601; 32101; 88259 88259 32101; 88259;44601 88259 11342; 88259 11342; 88259 11342; 25107 89349; 18117; 94195; 34871 89349; 89053 298 72807 93981 7572; 13524; 19497 Caldera name Campi Flegrei Hatori Narugo Mukaimachi Onikobe Towada Okiura Hakkoda Toya Shikotsu Akaigawa Tokachi Daisetsu Sumisu Myojinsho Myojin Knoll

240

5 9 2 2 5 2 9 5 5 8 11 11 11 11 11 11 11 Available # Parameters 1 1 2 2 2 2 1 2 1 1 1 1 1 1 1 1 1 Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 1 1 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 Crust Classification: 1 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 Classification: Tectonics setting Tectonics References 52414 46307 89524 23212; 24053 94049 72245 136; 98120; 68362; 94225 108; 28082; 136; 45303; 451; 31852; 7496; 6095; 25832 287 50421 25832 46811; 82126; 35388 87427 25832 6095; 25832 46811; 242; 46811; 25927 242; 46811; 46811; Caldera name Higashi Kurose Hole Kakuto Ata Aira Aso Shishimuta Kikai Hangar Pauzhetka Ksudach Opala Gorely Akademii Karymshina Stena-Soboliny Bolshoi Semi.

241

2 2 5 5 2 5 5 5 5 5 6 11 11 11 11 11 11 Available # Parameters 1 1 2 1 1 2 2 2 1 1 1 1 1 1 1 1 1 Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 3 3 1 1 1 2 2 2 1 1 1 1 1 1 1 1 1 Crust Classification: 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Classification: Tectonics setting Tectonics References 85372; 7866 108; 136; 63969; 7518 58433-EarthChem 80822 62000 39606; 89053 1314 (Supplement); 26511 15437 14029; 73618-Earth Chem 14029 15437; 38390 14029 15437 14029 14029; 35459 Caldera name Uzon Krasheninnikov Karymsky Macauley Denham Akan Kutcharo L'vinaya Past Golovnin Mendeleev Urbich Isthmus Vetrovoi Medvezhii Chirpoi Zavaritsky Tao-Rusyr Nemo

242

5 2 10 11 11 11 11 11 11 11 11 11 10 10 10 11 11 Available # Parameters 1 2 2 2 2 2 2 2 2 2 2 1 1 1 1 2 1 Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 1 1 3 3 3 3 3 3 3 3 3 1 2 2 2 2 3 Crust Classification: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 1 Classification: Tectonics setting Tectonics References 26054; 53610 All info from Smith paper in progress: 59 9393 65852 9604 58071 44837; 75376 21789 11844 74913; 75089 25832 1498; 26631 78901 1178 1178 136l; 83893; 88; 1934; 89755 3820 26405 Caldera name Pitons Wotten Waven Wotten Trois Morne Humeros, Los Primavera, La Taupo Whakamaru Reporoa Mangakino Okataina Rotorua Kapenga El Valle Witori-Pago Rabaul Tavui Lolobau Laguna de Bay

243

5 5 11 11 10 11 11 10 11 Available # Parameters 2 1 2 2 2 2 2 2 1 Classification: Stress regime (1 = regime Stress compressional; 2 = compressional; 2 2 3 3 3 3 3 3 1 Crust Classification: 1 1 3 3 3 3 3 1 3 Classification: Tectonics setting Tectonics References 26511 30889 108; 94922; 136; 34318 136; 44673 136; 10747; 56985; 39277 108; 96977; 58200 136; 6556; 68284; 15; 51068; 67103; 136; 65; 89829; 2718 8200 72191 Caldera name Calderas used to test geochem (not included in main database) References are given in the main text, Table 6 given in the main text, Table are References Taal Irosin Deriba Acigol Nemrut Dagi Long Valley Caldera Valles Yellowstone Kuwae

244

2 2 2 2 2 2 65.4 74.82 77.53 67.56 68.61 72.90 SiO2 (wt%) Stress regime Classification: 3 3 3 3 3 3 ? ? 65.4 77.5 75.22 67.56 Crust (wt%) Classification: High SiO2 CFE 2 1 2 2 1 1 1 3 1 1 ? ? regime setting Tectonic Tectonic Local stress Classification: 9 4.2 373 100 33.8 0.37 (Ma) Date of Formation References See Table 6 Table See 7 Table See 8 Table See 9 Table See 10 Table See 11 Table See 5 15 27 21 8.5 9.5 -1000 ~0.705 0.70549 0.71027 0.70388 87/86 Sr Average Average Diameter 0.708-0.711 3 1 3 0 2 4 41 340 595 129 CFE 1015 1036 Ba (ppm) composition 3 3 3 1 3 3 31 13 299 186 136 25.8 type Crust Sr (ppm) 59 122 151 306 166 131 Lon Rb (ppm) -99.655 -66.2 –116.621 -110.95 145.7 -118.49 39 69 58 28 Lat 237 58.4 (ppm) Y + Nb Y 20.367 -24.5 40.156 18.78 -36.7 37.83 1 3 3 2 3 1 68 208 149 236 1353 125.00 setting Tectonic Tectonic Zr (ppm) 8.54 7.09 7.47 6.12 9.00 10.73 (wt%) Na2O + K2O Andes Belt, SE Australia Belt, SE Location Central Mexico Puna - Central Nevada Socorro Island Lachlan Fold Central Sierra Nevada, CA, USA Mexico Ignimbrite Ignimbrite Caldera name Huichipan - Don Guinyó Acay Negra Muerta caldera - Caetano Socorro Volcanics Town Violet Minarets Caldera Caldera name Huichipan - Don Guinyó Acay Negra Muerta caldera - Caetano Socorro Volcanics Town Violet Minarets Caldera Ignimbrite Ignimbrite

245

9 Appendix C

References for data listed in Appendices A and B

Reference Citation ID Number

4 Bindeman, I.N., Fournelle, J.H., and Valley, J.W., 2001, Low-delta O18 tephra from a compositionally zoned magma body; Fisher Caldera, Unimak Island, Aleutians: Journal of Volcanology and Geothermal Research v. 111, p. 35–53.

7 Dreher, S.T., 2002, The physical volcanology and petrology of the 3400 YBP caldera-forming eruption of Aniakchak Volcano, Alaska: v. PhD, p. 179.

9 Hildreth, W.H., 1996, Kulshan Caldera: A Quaternary subglacial caldera in the North Cascades, Washington: Geological Society of America Bulletin v. 108, p. 786–793.

12 Mahood, G.A., 1981, A summary of the geology and petrology of the Sierra la Primavera, Jalisco, Mexico; Granites and rhyolites: Journal of Geophysical Research, B v. 86, p. 10137–10152.

18 Templeton, J.H., 2004, Petrologic analysis of the Tepee Draw Tuff, Newberry Volcano, Oregon; preliminary constraints on compositional zoning in the pre- eruptive magma chamber: Geological Society of America, Rocky Mountain Section, 56th annual meeting v. 36(4), p. 8.

20 Reinecker, J., Heidbach, O., Tingay, M., Connolly, P., and Müller, B., 2004, World Stress Map, 2004 release.

21 Orsi, G., De, V., S., and Di, V., M., 1996, The restless, resurgent Campi Flegrei nested caldera (Italy): constraints on its evolution and configuration: Journal of Volcanology and Geothermal Research v. 74, p. 179–214.

24 Mann, D., and Freymueller, J., 2003, Volcanic and tectonic deformation on Unimak Island in the Aleutian Arc, Alaska: Journal of Geophysical Research, B, Solid Earth and Planets v. 108, p. 12.

37 Mangan, M.T., Waythomas, C.F., Miller, T.P., and Trusdell, F.A., 2003, Emmons Lake volcanic center, Alaska Peninsula; source of the late Wisconsin Dawson Tephra, Yukon Territory, Canada: Canadian Journal of Earth Sciences v. 40, p. 925–936.

46 Holbrook, W.S., Lizarralde, D., McGeary, S., Bangs, N., and Diebold, J., 1999, Structure and composition of the Aleutian island arc and implications for continental crustal growth: Geology v. 27, p. 31–34.

52 Campos-Enriquez, J.O., Dominguez-Mendez, F., Lozada-Zumaeta, A., Morales- Rodriguez, H.F., and Andaverde-Arredondo, J.A., 2005, Application of the Gauss theorem to the study of silicic calderas: The calderas of La Primavera, Los

246

Azufres, and Los Humeros (Mexico): Journal of Volcanology and Geothermal Research v. 147, p. 39–67.

53 Campos-Enriquez, J.O., Arroyo-Esquivel, M.A., and Urrutia-Fucugauchi, J., 1990, Basement, Curie isotherm and shallow-crustal structure of the Trans-Mexican volcanic belt, from aeromagnetic data: Tectonophysics v. 172, p. 77–90.

58 Ferriz, H., and Mahood, G.A., 1984, Eruption rates and compositional trends at Los Humeros volcanic center, Puebla, Mexico: Symposium on Calderas and related volcanic rocks; the Krakatau centennial; a part of the American Geophysical Union, 1983 fall meeting, San Francisco, CA, United States, Dec. 5-9, 1983 v. 89, p. 8511.

59 Ferriz, H., and Mahood, G.A., 1987, Strong compositional zonation in a silicic magmatic system: Los Humeros, Mexican neovolcanic belt: Journal of Petrology v. 28, p. 171–209.

60 Mahood, G.A., 1980, Geological evolution of a Pleistocene rhyolitic center; Sierra La Primavera, Jalisco, Mexico: Journal of Volcanology and Geothermal Research v. 8, p. 199–230.

61 Mahood, G.A., Truesdell, A.H., and Templos M, L.A., 1983, A reconnaissance geochemical study of La Primavera geothermal area, Jalisco, Mexico: Journal of Volcanology and Geothermal Research v. 16, p. 247–261.

64 MacLeod, N.S., Jr., and Sherrod, D.R., 1988, Geologic evidence for a magma chamber beneath Newberry Volcano, Oregon;: Journal of Geophysical Research, B v. 93, p. 10067–10079.

65 Peng, X., and Humphreys, E.D., 1998, Crustal velocity structure across the eastern Snake River plain and the Yellowstone Swell: Journal of Geophysical Research, B, Solid Earth and Planets v. 103, p. 7171.

69 Ferrés, D., Delgado, H., Hernández, W., and Gutiérrez, E., 2007, Destruction of San Salvador Volcano and Birth of El Boqueròn Volcano (El Salvador): Detailed Stratigraphic Study of G1 and G2 Sequences: EOS Transactions, AGU, v. 88 Abstract V31B-0490.

72 Mooney, W.D., and Weaver, C.S., 1989, Regional crustal structure and tectonics of the Pacific coastal states; California, Oregon, and Washington Geophysical framework of the continental United States: Geological Society of America Memoir v. 172, p. 129–161.

78 Fliedner, M.M., and Klemperer, S.L., 2000, Crustal structure transition from oceanic arc to continental arc, eastern Aleutian Islands and Alaska Peninsula: Earth and Planetary Science Letters v. 179, p. 567–579.

79 Lizarralde, D., Holbrook, W.S., McGeary, S., Bangs, N.L., and Diebold, J.B., 2002, Crustal construction of a volcanic arc, wide-angle seismic results from the western Alaska Peninsula: Journal of Geophysical Research, B v. 107, p. EPM 4–1 - 21.

88 Bailey, R.A., 1989, Geologic map of Long Valley Caldera, Mono-Inyo Craters volcanic chain, and vicinity, eastern California: v. 11.

94 Bassin, C., Laske, G., and Masters, G., 2000, The current limits of resolution for surface wave tomography in North America: EOS Transactions, American

247

Geophysical Union v. 81, p. 897.

98 Hildreth, W., Lanphere, M.A., and Fierstein, J., 2003, Geochronology and eruptive history of the Katmai volcanic cluster, Alaska Peninsula: Earth and Planetary Science Letters v. 214, p. 93–114.

108 Newhall, C.G., and Dzurisin, D., 1988, Historical unrest at large calderas of the world, Volumes 1 and 2: US Geological Survey Bulletin v. 1855, p. 1108.

135 Rose, W.I., Conway, F.M., Pullinger, C.R., Deino, A., and MacIntosh, W.C., 1999, An improved age framework for late Quaternary silicic eruptions in northern Central America: Bulletin of Volcanology v. 61, p. 106–120.

136 Siebert, L., and Simkin, T., 2002, Volcanoes of the World: an Illustrated Catalog of Holocene Volcanoes and their Eruptions: Smithsonian Institution,Global Volcanism Program Digital Information Series, GVP-3, http://www.volcano.si.edu/world/ v. Last accessed February 1, 2009.

141 Siebel, W., Schnurr, W.B.W., Hahne, K., Kraemer, B., Trumbull, R.B., van den Bogaard, P., and Emmermann, R., 2001, Geochemistry and isotope systematics of small- to medium-volume Neogene-Quaternary ignimbrites in the southern Central Andes: Evidence for derivation from andesitic magma sources: Chemical Geology v. 171, p. 213–237.

153 Silberling, N.J., Jones, D.L., Monger, J.W.H., and Coney, P.J., 1992, Lithotectonic terrane map of the North American Cordillera: U. S. Geological Survey, Report 1- 2176 v. 2 sheets.

159 Druitt, T.H., Bacon, C.R., and Brown, P.E., 1988, Compositional zonation and cumulus processes in the Mount Mazama magma chamber, Crater Lake, Oregon: Transactions of the Royal Society of Edinburgh, Earth Sciences v. 79, p. 289–297.

166 Beck, S.L., and Zandt, G., 2002, The nature of orogenic crust in the Central Andes: Journal of Geophysical Research, B v. 107, p. 2230–2247.

167 Bohm, M., Lueth, S., Echtler, H., Asch, G., Bataille, K., Bruhn, C., Rietbrock, A., and Wigger, P., 2002, The Southern Andes between 36° and 40°S latitude: Seismicity and average seismic velocities: Tectonophysics v. 356, p. 275–289.

168 Vincent, P.M., 1970, The evolution of the Tibesti volcanic province, eastern Sahara: African Magmatism and Tectonics, Oliver and Boyd, Edinbourg v. 301– 309.

194 Newhall, C.G., 1987, Geology of the Lake Atitlan region, western Guatemala: Journal of Volcanology and Geothermal Research v. 33, p. 23–55.

198 Ramos, V.A., and Aleman, A.M., 2000, Tectonic evolution of the Andes, in Cordani, U.G., Milani, E.J., and Thomaz Filho, A., ed., Tectonic evolution of South America, p. 635–685.

211 Mason, B., Lipman, P., Self, S., Ponomareva, V., Smellie, J., Permenter, J., Fridrich, C., Allen, S., Isaia, R., Wu, J., and Pyle, D., 2005, Volcanic eruption database:

212 Kempter, K.A., Benner, S.G., and Williams, S.N., 1996, Rincon de la Vieja Volcano, Guanacaste Province, Costa Rica: Geology of the southwestern flank and hazards implications: Journal of Volcanology and Geothermal Research v.

248

71, p. 109–127.

215 Bellier, O., Bellon, H., Sebrier, M., Sutanto, and Maury, R.C., 1999, K-Ar age of the Ranau tuffs: implications for the Ranau Caldera emplacement and slip-partitioning in Sumatra (Indonesia): Tectonophysics v. 312, p. 347–359.

234 Bogoyavlenskaya, G.E., 1973, Melted xenoliths of intrusive rocks in the pyroclastic deposits of the Uzon and Semyatchinskaya ring structures (Kamchatka): Bulletin Volcanologique v. 36, p. 443–445.

238 Bindeman, I.N., Vinogradov, V.I., Valley, J.W., Wooden, J.L., and Natalʼin, B.A., 2002, Archean protolith and accretion of crust in Kamchatka; SHRIMP dating of zircons from Sredinny and Ganal massifs: Journal of Geology v. 110, p. 271–289.

242 Erlich, E.N., 1986, Geology of the calderas of Kamchatka and Kurile Islands with comparison to calderas of Japan and the Aleutians, Alaska: US Geological Survey Open File Report 86-0291, 300 p.

244 Hazlett, R.W., 1987, Geology of the San Cristobal volcanic complex, Nicaragua: Journal of Volcanology and Geothermal Research v. 33, p. 223–230.

245 Izbekov, P., Gardner, J.E., and Eichelberger, J.C., 2004, Comagmatic granophyre and dacite from Karymsky volcanic center, Kamchatka; experimental constraints for magma storage conditions: Journal of Volcanology and Geothermal Research v. 131, p. 1–18.

246 Levin, V., Park, J., Brandon, M., Lees, J.M., Peyton, V., Gordeev, E., and Ozerov, A., 2002, Crust and upper mantle of Kamchatka from teleseismic receiver functions; Structure of the continental lithosphere and upper mantle: Tectonophysics v. 358, p. 233–265.

261 Mann, C.P., Stix, J., Vallance, J.W., and Richer, M., 2004, Subaqueous intracaldera volcanism, Ilopango Caldera, El Salvador, Central America, in Rose, W.I., Bommer, J.J., Lopez, D.L., Carr, M.J., and Major, J.J., ed., Natural Hazards in El Salvador, GSA Special Paper 375: Boulder, CO, Geological Society of America, p. 159–174.

263 Parini, M., Pisani, P., and Monterrosa, M., 1995, Resource Assessment at the Berlin Geothermal Field (El Salvador): Proceedings of the World Geothermal Congress v. 3, p. 1537–1548.

267 Rotolo, S.G., and Castorina, F., 1998, Transition from mildly-tholeiitic to calc- alkaline suite: the case of Chichontepec volcanic centre, El Salvador, Central America: Journal of Volcanology and Geothermal Research v. 86, p. 117–136.

273 Sofield, D., 2004, Eruptive history and volcanic hazards of Volcan San Salvador r, in Rose, W.I., Bommer, J.J., Lopez, D.L., Carr, M.J., and Major, J.J., ed., Natural hazards in El Salvador: GeoEngineers, Tacoma, WA, United States (USA); Imperial College London, United Kingdom (GBR); Ohio University, United States (USA); U. S. Geological Survey, United States (USA); Michigan Technological University, Department of Geological Enginee(TRUNCATED) United States (USA), Geological Society of America (GSA), Boulder, CO, United States (USA), p. 147–158.

278 Tabor, R.W., Haugerud, R.A., Hildreth, W., and Brown, E.H., 2003, Geologic map of the Mount Baker 30- by 60-minute quadrangle, Washington: US Geological

249

Survey, Report I-2660 v. 2 sheets, 74 p.

279 Taira, A., 2001, Tectonic evolution of the Japanese Island arc system: Annual Review of Earth and Planetary Sciences v. 29, p. 109–134.

285 Vallier, T.L., Scholl, D.W., Fisher, M.A., Bruns, T.R., Wilson, F.H., von Huene, R., Stevenson, A.J., Plafker, G., and Berg, H.C., 1994, Geologic framework of the Aleutian Arc, Alaska, in Plafker, G., and Berg, H.C., ed., The Geology of Alaska: Boulder, CO USA, Geological Society of America, p. 367–388.

286 Verma, S.P., 2000, Geochemical evidence for a lithospheric source for magmas from Los Humeros Caldera, Puebla, Mexico: Chemical Geology v. 164, p. 35–60.

287 Volynets, O.N., Ponomareva, V.V., Braitseva, O.A., Melekestsev, I.V., and Chen, C.H., 1999, Holocene eruptive history of Ksudach volcanic massif, South Kamchatka; evolution of a large magmatic chamber: Journal of Volcanology and Geothermal Research v. 91, p. 23–42.

298 Tamura, Y., and Tatsumi, Y., 2002, Remelting of an Andesitic crust as a possible origin for rhyolitic magma in oceanic arcs: an example from the Izu--Bonin Arc: Journal of Petrology v. 43, p. 1029–1047.

318 Kerle, N., 2001, The 1998 debris avalanche at Casita volcano, Nicaragua- investigation of structural deformation as the cause of slope instability using remote sensing: Journal of Volcanology and Geothermal Research v. 105, p. 49– 63.

324 Konstantinovskaia, E.A., 2000, Geodynamics of an Early Eocene arc-continent collision reconstructed from the Kamchatka Orogenic Belt, NE Russia: Tectonophysics v. 325, p. 87–105.

407 Blake, D.H., 1976, Pumiceous pyroclastic deposits of Witori Volcano, New Britain, Papua New Guinea: Volcanism in Australasia: A Collection of Papers in Honour of the Late GAM Taylor v. 191–200.

451 Ponomareva, V., Churikova, T., Melekestsev, I., Braitseva, O., Pevzner, M., and Sulerzhitsky, L., 2007, Late Pleistocene-Holocene Volcanism on the Kamchatka Peninsula, Northwest Pacific Region: Geophysical Monograph v. 172, p. 165.

873 DeMoor, M., Varekamp, J., Goss, A., Merrill, M., Colvin, A., and Vroon, P., 2004, Magmatic evolution of the Copahue-Caviahue volcanic complex: Rocky Mountain and Cordilleran Joint Meeting v. 36(4), p. 69.

1178 Wood, C.P., Nairn, I.A., Mckee, C.O., and Talai, B., 1995, Petrology of the Rabaul Caldera area, Papua New Guinea: Journal of Volcanology and Geothermal Research v. 69, p. 285–302.

1306 Tamura, Y., Gill, J.B., Tollstrup, D., Kawabata, H., Shukuno, H., Chang, Q., Miyazaki, T., Takahashi, T., Hirahara, Y., Kodaira, S., and others, 2009, Silicic Magmas in the Izu-Bonin Oceanic Arc and Implications for Crustal Evolution: Journal of Petrology v. 50, p. 685.

1314 Vogel, T.A., Flood, T.P., Patino, L.C., Wilmot, M.S., Maximo, R.P.R., Arpa, C.B., Arcilla, C.A., and Stimac, J.A., 2006, Geochemistry of silicic magmas in the Macolod Corridor, SW Luzon, Philippines: evidence of distinct, mantle-derived, crustal sources for silicic magmas: Contributions to Mineralogy and Petrology v.

250

151, p. 267–281.

1340 Wohletz, K.H., 1986, Explosive magma-water interactions: Thermodynamics, explosion mechanisms, and field studies: Bulletin of Volcanology v. 48, p. 245– 264.

1498 Defant, M.J., Clark, L.F., Stewart, R.H., Drummond, M.S., de, B., J.Z., Maury, R.C., Bellon, H., Jackson, T.E., and Restrepo, J.F., 1991, Andesite and dacite genesis via contrasting processes: the geology and geochemistry of El Valle Volcano, Panama: Contributions to Mineralogy and Petrology v. 106, p. 309–324.

1618 Hildreth, W., Halliday, A.N., and Christiansen, R.L., 1991, Isotopic and chemical evidence concerning the genesis and contamination of basaltic and rhyolitic magma beneath the Yellowstone Plateau volcanic field: Journal of Petrology v. 32, p. 63–138.

2202 Miller, T. P., McGimsey, R. G., Richter, D.H., Riehle, J.R., Nye, C.J., Yount, M.E., and Dumoulin, J.A., Catalog of the historically active volcanoes of Alaska: USGS Open-File Report 98-582, 104 p.

2481 Mu-oz, B.J., and Stern, C.R., 1988, The Quaternary volcanic belt of the southern continental margin of South America: transverse structural and petrochemical variations across the segment between 38 S and 39 S: Journal of South American earth sciences v. 1, p. 147–161.

2950 Tunstall, C., and Folguera, A., 2005, Structural control on the development of an anomalous concentration of calderas in the Neuquén Andes: Pino Hachado volcanic complex (38° 30ʼ S-71° W): Revista de la Asociación Geológica Argentina v. 60, p. 731–741.

3117 Shinjo, R., Woodhead, J.D., and Hergt, J.M., 2000, Geochemical variation within the northern Ryukyu Arc: magma source compositions and geodynamic implications: Contributions to Mineralogy and Petrology v. 140, p. 263–282.

3193 Goss, A.R., Kay, S.M., Depine, G., and Mpodozis, C., 2006, The Incapillo Caldera: A stranded magma chamber over a dying Andean arc: EOS (Transactions, American Geophysical Union) v. 87 Fall Meeting Supplement, Abstract V11A- 0557,

3287 Varekamp, J.C., deMoor, J.M., Merrill, M., Colvin, A., Goss, A., Vroon, P., and Hilton, D.R., 2006, Geochemistry and isotopic characteristics of the Caviahue- Copahue volcanic complex, Province of Neuquen, Argentina: Evolution of an Andean Margin, Geological Society of America Special Paper v. 407, p. 317–342.

3385 Kimura, J.I., and Yoshida, T., 1999, Mantle diapir-induced arc volcanism: The Ueno Basalts, Nomugi-Toge and Hida volcanic suites, central Japan: The Island Arc v. 8, p. 304–322.

3501 Deering, C.D., Vogel, T.A., Patino, L.C., and Alvarado, G.E., 2007, Origin of distinct silicic magma types from the Guachipelín Caldera, NW Costa Rica: Evidence for magma mixing and protracted subvolcanic residence: Journal of Volcanology and Geothermal Research v. 165, p. 103–126.

3589 Wang, Y., Li, C., Wei, H., and Shan, X., 2003, Late Pliocene--recent tectonic setting for the Tianchi volcanic zone, Changbai Mountains, northeast China: Journal of Asian Earth Sciences v. 21, p. 1159–1170.

251

3820 Smith, I.E.M., and Johnson, R.W., 1981, Contrasting rhyolite suites in the late cenozoic of Papua New Guinea: Journal of Geophysical Research v. 86, p. 10257–10272.

4183 Reubi, O., and Nicholls, I.A., 2004, Variability in eruptive dynamics associated with caldera collapse: an example from two successive eruptions at Batur volcanic field, Bali, Indonesia: Bulletin of Volcanology v. 66, p. 134–148.

4295 Takahashi, N., Uyehiro, K.S., and Hinohara, M.S., 1998, Implications from the seismic crustal structure of the northern Izu-Bonin arc: The Island Arc v. 7, p. 383–394.

4298 Tregoning, P., Lambeck, K., Stolz, A., Morgan, P., McClusky, S.C., van, d.B., P., McQueen, H., Jackson, R.J., Little, R.P., Laing, A., and others., 1998, Estimation of current plate motions in Papua New Guinea from Global Positioning System observations: Journal of Geophysical Research v. 103

4308 Chevallier, L., and Verwoerd, W.J., 1988, A numerical model for the mechanical behavior of intraplate volcanoes: Journal of Geophysical Research, B v. 93, p. 4182–4198.

4538 Bacon, C.R., Mastin, L.G., Scott, K.M., and Nathenson, M., 1997, Volcano and earthquake hazards in the Crater Lake region, Oregon: US Geological Survey Open File Report 97-0487, 32 p.

4725 Kiriyanov, V.Y., Melekestsev, I.V., and Felitsyn, S.B., 1994, The eruption of Opala Volcano in 500 A.D.: One of the largest explosive eruptions in Kamchatka in the last 2,000 years: EOS (Transactions, American Geophysical Union) v. 75, p. 732.

5147 Aleman, A., and Ramos, V.A., 2000, Northern Andes, in Cordani, U.G., Milani, E.J., Thomaz Filho, A., and Campos, D.A., ed., Tectonic evolution of South America, p. 453–480.

5184 Yoshida, H., and Takahashi, M., 1991, Geology of the eastern part of the Shirakawa pyroclastic flow field: Journal of the Geological Society of Japan v. 97, p. 231–249.

5240 Delong, S.E., Perfit, M.R., McCulloch, M.T., and Ach, J., 1985, Magmatic evolution of Semisopochnoi Island, Alaska: trace-element and isotopic constraints: The Journal of Geology v. 609–618.

5305 Sepulveda, F., Lahsen, A., Bonvalot, S., Cembrano, J., Alvarado, A., and Leteller, P., 2005, Morpho-structural evolution of the Cordón Caulle geothermal region, Southern Volcanic Zone, Chile: Insights from gravity and 40Ar / 39Ar dating: Journal of Volcanology and Geothermal Research v. 148, p. 165–189.

5513 Yamagata, K., 1991, Formation of lithic breccia and vent evolution during the 32 KA. Eruption of Shikotsu caldera, Japan: Geographical Reports of Tokyo Metropolitan University v. 26, p. 227–240.

5537 Vogel, T.A., Patino, L.C., Alvarado, G.E., and Gans, P.B., 2004, Silicic ignimbrites within the Costa Rican volcanic front; evidence for the formation of continental crust: Earth and Planetary Science Letters v. 226, p. 149–159.

5590 Katili, J.A., 1975, Volcanism and plate tectonics in the Indonesian island arcs: Tectonophysics v. 26, p. 165–188.

252

5824 Riller, U., Petrinovic, I., Ramelow, J., Strecker, M.R., and Oncken, O., 2001, Late Cenozoic tectonism, collapse caldera and plateau formation in the Central Andes: Earth and Planetary Science Letters v. 188, p. 299–311.

5913 Barberi, F., Coltelli, M., Ferrara, G., Innocenti, F., Navarro, J.M., and Santacroce, R., 1988, Plio-Quaternary volcanism in Ecuador: Geological Magazine v. 125, p. 1–14.

6095 Bindeman, I.N., Leonov, V.L., Izbekov, P.E., Ponomareva, VV, Watts, K.E., Shipley, N.K., Perepelov, A.B., Bazanova LI, Jicha BR, Singer BS, Schmitt AK, Portnyagin MV, and Chen, C.H., 2009, Large-volume silicic volcanism in Kamchatka: Ar-Ar and U-Pb ages, isotopic, and geochemical characteristics of major pre-Holocene caldera-forming eruptions: Journal of Volcanology and Geothermal Research v. in press

6225 Wiart, P., and Oppenheimer, C., 2005, Large magnitude silicic volcanism in north Afar: the Nabro Volcanic Range and Maʼalalta volcano: Bulletin of Volcanology v. 67, p. 99–115.

6556 Heiken, G., Goff, F., Gardner, J.N., Baldridge, W.S., Hulen, J.B., Nielson, D.L., and Vaniman, D., 1990, The Valles/Toledo caldera complex, Jemez volcanic field, New Mexico: Annual Review of Earth and Planetary Sciences v. 18, p. 27–53.

6681 Catchings, R.D., and Mooney, W.D., 1988, Crustal structure of east central Oregon: Relation between Newberry Volcano and regional crustal structure: Journal of Geophysical Research, B v. 93, p. 10081–10094.

7496 Ponomareva, V.V., Kyle, P.R., Melekestsev, I.V., Rinkleff, P.G., Dirksen, O.V., Sulerzhitsky, L.D., Zaretskaia, N.E., and Rourke, R., 2004, The 7600 (14 C) year BP Kurile Lake caldera-forming eruption, Kamchatka, Russia: stratigraphy and field relationships: Journal of Volcanology and Geothermal Research v. 136, p. 199–222.

7518 Darbyshire, F.A., White, R.S., and Priestley, K.F., 2000, Structure of the crust and uppermost mantle of Iceland from a combined seismic and gravity study: Earth and Planetary Science Letters v. 181, p. 409–428.

7522 Newhall, C.G., Paull, C.K., Bradbury, J.P., Higuera-Gundy, A., Poppe, L.J., Self, S., Sharpless, N.B., and Ziagos, J., 1987, Recent geologic history of lake Atitlan, a caldera lake in western Guatemala: Journal of Volcanology and Geothermal Research v. 33, p. 81–107.

7572 Civetta, L., Orsi, G., Pappalardo, L., Fisher, R.V., Heiken, G., and Ort, M., 1997, Geochemical zoning, mingling, eruptive dynamics and depositional processes— the Campanian Ignimbrite, Campi Flegrei caldera, Italy: Journal of Volcanology and Geothermal Research v. 75, p. 183–219.

7866 Zolotarev, B.P., Karpov, G.A., Yeroshev-Shak, V.A., Artamonov, A.V., Grigoryev, V.S., and Pokrovsky, B.G., 1999, Evolution of volcanism in Uzon caldera: Volcanology and Seismology v. 6, p. 67–84.

8200 Robin, C., Monzier, M., and Eissen, J.P., 1994, Formation of the mid-fifteenth century Kuwae caldera (Vanuatu) by an initial hydroclastic and subsequent ignimbritic eruption: Bulletin of Volcanology v. 56, p. 170–183.

253

8259 Bacon, C.R., 1983, Eruptive history of Mount Mazama and Crater Lake caldera, Cascade Range: Journal of Volcanology and Geothermal Research v. 18, p. 57– 115.

8318 Vincent, Pierre M, Gourgaud, Alain, and Mbaitoudji, M., 1994, Le Trou au Natron du Tousside (Tibesti, Tchad); une caldera polygenique complexe. Trou au Natron (Tousside, Tibesti, Chad); a complex polygenetic caldera: Reunion Annuelle des Sciences de la Terre v. 15, p. 66.

8869 Hallinan, S., and Brown, G., 1995, Incremental collapse and stratocone growth within a funnel-shaped caldera, Guayabo, Costa Rica: Journal of Volcanology and Geothermal Research v. 67, p. 101–122.

8920 Machida, H., 1999, The stratigraphy, chronology and distribution of distal marker- tephras in and around Japan: Global and Planetary Change v. 21, p. 71–94.

9393 Mahood, G.A., and Halliday, A.N., 1988, Generation of high-silica rhyolite: A Nd, Sr, and O isotopic study of Sierra La Primavera, Mexican Neovolcanic Belt: Contributions to Mineralogy and Petrology v. 100, p. 183–191.

9604 Sutton, A.N., Blake, S., and Wilson, C.J.N., 1995, An outline geochemistry of rhyolite eruptives from Taupo volcanic centre: New Zealand. Journal of Volcanology and Geothermal Research v. 68, p. 153–175.

9700 Yumul, G.P., Dimalanta, C., Bellon, H., Faustino, D.V., De Jesus, J.V., Tamayo, R.A., and FT, J., 2000, Adakitic lavas in the Central Luzon back-arc region, Philippines: lower crust partial melting products?: The Island Arc v. 9, p. 499–512.

9880 Leat, P.T., Macdonald, R., and Smith, R.L., 1984, Geochemical evolution of the Menengai caldera volcano, Kenya: Journal of Geophysical Research v. 89

9941 Sparks, R.S.J., and Sigurdsson, H., 1987, The 1875 eruption of Askja volcano, Iceland: combined fractional crystallization and selective contamination in the generation of rhyolitic magma: Mineralogical Magazine v. 51, p. 183–202.

10071 Kay, S.M., Godoy, E., and Kurtz, A., 2005, Episodic arc migration, crustal thickening, subduction erosion, and magmatism in the south-central Andes: Bulletin of the Geological Society of America v. 117, p. 67–88.

10350 Carter, L., Shane, P., Alloway, B., Hall, I.R., Harris, S.E., and Westgate, J.A., 2003, Demise of one volcanic zone and birth of another-A 12 my marine record of major rhyolitic eruptions from New Zealand: Geology v. 31, p. 493-496.

10407 Lindley, D., 1988, Early Cainozoic stratigraphy and structure of the Gazelle Peninsula, east New Britain: An example of extensional tectonics in the New Britain arc-trench complex: Australian Journal of Earth Sciences v. 35, p. 231– 244.

10643 Ostapenko, V.F., Fedorchenko, V.I., and Shilov, V.N., 1967, Pumices, ignimbrites and rhyolites from the Great Kurile Arc: Bulletin of Volcanology v. 30, p. 81–92.

10747 Ulusoy, I., Labazuy, P., Aydar, E., Ersoy, O., and Çubukçu, E., 2008, Structure of the Nemrut caldera (Eastern Anatolia, Turkey) and associated hydrothermal fluid circulation: Journal of Volcanology and Geothermal Research v. 174, p. 269–283.

254

10863 Geologi, P.P.d.P., 2003, Atlas, Geology and Potency of Mineral and Energy Resources, Indonesian Region, scale 1:10,000,000: Bandung, Indonesia, Pusat Penelitian dan Pengembangan Geologi, 39 p.

11342 Kudo, T., Sasaki, M., Uchiyama, Y., Nozawa, A., Sasaki, H., Tokizawa, T., and Takarada, S., 2007, Petrological variation of large-volume felsic magmas from Hakkoda-Towada caldera cluster: Implications for the origin of high-K felsic magmas in the Northeast Japan Arc: The Island Arc v. 16, p. 133–155.

11477 Takada, A., Nasution, A., and Mulyana, R., 2003, Eruptive history during the last 10 ky for the caldera-forming eruption of Rinjani volcano, Indonesia: Japan Earth and Planetary Science Joint Meeting v. Abstract V055–001.

11521 Samaniego, P., Martin, H., Monzier, M., Robin, C., Fornari, M., Eissen, J.-P., and Cotten, J., 2005, Temporal evolution of magmatism in the northern volcanic zone of the Andes: the geology and petrology of Cayambe volcanic complex (Ecuador): Journal of Petrology v. 46, p. 2225–2252.

11844 Beresford, S.W., and Cole, J.W., 2000, Kawerau Ignimbrite: a 0.24 Ma ignimbrite erupted from the Okataina caldera complex, Taupo Volcanic Zone, New Zealand: New Zealand Journal of Geology and Geophysics v. 43, p. 109–116.

11867 Melekestsev, I. V., Ponomareva, V. V., and Volynets, O.N., 1995, Kizimen Volcano, Kamchatka--A future Mount St. Helens?: Journal of Volcanology and Geothermal Research v. 65, p. 205–226.

12199 Chiesa, S., Civelli, G., Gillot, P.Y., Mora, О., and Alvarado, G., 1992, Rocas piroclásticas asociadas con la formación de la Caldera de Guayabo, Cordillera de Guanacaste, Costa Rica: Revista Geolica de América Central v. 14, p. 59–75.

12245 Braitseva, O.A., and Melekestsev, I.V., 1991, Eruptive history of Karymsky volcano, Kamchatka, USSR, based on tephra stratigraphy and 14 C dating: Bulletin of Volcanology v. 53, p. 195–206.

12288 Stelling, P., Gardner, J.E., and Begèt, J., 2005, Eruptive history of Fisher Caldera, Alaska, USA: Journal of Volcanology and Geothermal Research v. 139(3-4), p. 163–183.

12894 Audley-Charles, M.G., 1991, Tectonics of the New Guinea Area: Annual Reviews in Earth and Planetary Sciences v. 19, p. 17–41.

13524 Rapolla, A., Fedi, M., and Fiume, M.G., 1989, Crustal Structure of the Ischia- Phlegrean Geothermal Fields, Near Naples, Italy, From Gravity and Aeromagnetic Datat: Geophysical Journal International v. 97, p. 409–419.

13591 Machida, H., Blong, R.J., Specht, J., Moriwaki, H., Torrence, R., Hayakawa, Y., Talai, B., Lolok, D., and Pain, C.F., 1996, Holocene explosive eruptions of Witori and Dakataua caldera volcanoes in West New Britain, Papua New Guinea: Quaternary International v. 34, p. 65–78.

13779 Goss, A.R., 2007, Personal communication: Stress regime around Incapillo:

13887 Nishimura, S., Harjono, H., and Suparka, S., 1992, The Krakatau islands: The geotectonic setting: GeoJournal v. 28, p. 87–98.

255

13923 Rose, W.I., Newhall, C.G., Bornhorst, T.J., and Self, S., 1987, Quaternary silicic pyroclastic deposits of Atitlan caldera, Guatemala: Journal of Volcanology and Geothermal Research v. 33, p. 57–80.

14029 Gorshkov, G.S., 1970, Volcanism and the upper mantle: investigations in the Kurile Island Arc: New York, Plenum Press, 385 p.

14249 Folguera, A., Ramos, V.A., Zapata, T., Spagnuolo, M., and Miranda, F., 2005, Pliocene to Quaternary retro-arc extension in the Andes at 35-37° 30ʼ S: 6th International Symposium on Andean Geodynamics, Extended Abstracts, p. 277– 280.

14470 Lara, LE, Moreno, H., Naranjo, J.A., Matthews, S., and Perez de Arce, C., 2006, Magmatic evolution of the Puyehue-Cordon Caulle Volcanic Complex (40oS), Southern Andean Volcanic Zone: From shield to unusual rhyolitic fissure volcanism: Journal of Volcanology and Geothermal Research v. 157, p. 343–366.

14930 Venable, M.E., 1994, A geologic, tectonic, an metallogenic evaluation of Siuna Terrane: Unpublished PhD thesis, University of Arizona, Tucson: 154 p.

14974 Stern, R.J., Fouch, M.J., and Klemperer, S.L., 2003, An overview of the Izu-Bonin- Mariana subduction factory, in Eiler, J., ed., Inside the Subduction Factory: p. 175–222.

15437 Otake, M., 2000, Stratigraphy and styles of caldera-forming eruption and subsidence of the Akakura caldera in the South Kurikoma geothermal area, Northeast Japan: Journal-Geological Society of Japan v. 106, p. 205–222.

15611 Saito, G., Stimac, J.A., Kawanabe, Y., and Goff, F., 2002, Mafic-felsic magma interaction at Satsuma-Iwojima volcano, Japan: Evidence from mafic inclusions in rhyolites: Earth Planets and Space v. 54, p. 303–326.

15714 Gunnarsson, B., Marsh, B.D., and Taylor, H.P., 1998, Generation of Icelandic rhyolites: silicic lavas from the Torfajökull central volcano: Journal of Volcanology and Geothermal Research v. 83, p. 1–45.

15730 Gerlach, D.C., Frey, F.A., Moreno-Roa, H., and Lopez-Escobar, L., 1988, Recent Volcanism in the Puyehue--Cordon Caulle Region, Southern Andes, Chile (40.5° S): Petrogenesis of Evolved Lavas: Journal of Petrology v. 29, p. 333–382.

15734 Pe-Piper, G., Piper, D.J.W., and Perissoratis, C., 2005, Neotectonics and the Kos Plateau Tuff eruption of 161 ka, South Aegean arc: Journal of Volcanology and Geothermal Research v. 139, p. 315–338.

15824 Milner, D., Cole, J., and Wood, C., 2002, Asymmetric, multiple-block collapse at Rotorua Caldera, Taupo Volcanic Zone, New Zealand: Bulletin of Volcanology v. 64, p. 134–149.

15836 Takahashi, M., and Yoshida, H., 1996, Evolution of magma plumbing system of the Shirakawa pyroclastic flow deposits: Journal of Mineralogy, Petrology and Economic Geology v. 91, p. 177–184.

16030 Endrun, B., 2005, Structural properties of the Hellenic subduction zone derived from receiver functions and surface wave dispersion: Geophysical Research Abstracts v. 7 #07123.

256

16555 Dull, R.A., Southon, J.R., and Sheets, P., 2001, Volcanism, ecology and culture: A reassessment of the Volcán Ilopango TBJ eruption in the southern Maya realm: Latin American Antiquity v. 25–44.

16666 Bailey, J.C., Frolova, T.I., and Burikova, I.A., 1989, Mineralogy, geochemistry and petrogenesis of Kurile island-arc basalts: Contributions to Mineralogy and Petrology v. 102, p. 265–280.

16841 Suzuki, T., Eden, D., Danhara, T., and Fujiwara, O., 2005, Correlation of the Hakkoda--Kokumoto Tephra, a widespread Middle Pleistocene tephra erupted from the Hakkoda Caldera, northeast Japan: The Island Arc v. 14, p. 666–678.

17164 Ballance, P.F., Ablaev, A.G., Pushchin, I.K., Pletnev, S.P., Birylina, M.G., Itaya, T., Follas, H.A., and Gibson, G.W., 1999, Morphology and history of the Kermadec trench-arc-backarc basin-remnant arc system at 30 to 32oS: geophysical profile, microfossil and K-Ar data: Marine Geology v. 159, p. 35–62.

17273 Harrington, R., 1989, The Diamante Caldera and Maipo Caldera complex in southern Andes of Argentina and Chile (24° 10ʼ S): Revista de la Asociacion Geologica Argentina v. 44, p. 186–193.

17486 Macdonald, R., Hawkesworth, C.J., and Heath, E., 2000, The Lesser Antilles volcanic chain: a study in arc magmatism: Earth-Science Reviews v. 49, p. 1–76.

18117 Hoe, S.G., Yamazaki, M., Sataka, H., Nakagawa, M., Akiyama, T., and Hirano, R., 2001, Tephrochronological reconsideration on Shikotsu Pumice Flow and Fall Deposits, Hokkaido, Japan.: Earth Science v. 55, p. 145–156.

18242 Kondo, H., Kudo, C., and Okamura, S., 2005, Chemical zonation of the rhyolitic Toya caldera magma body (Southwest Hokkaido, Japan): Asia Oceania Geosciences Society 2nd Annual Meeting v. Paper ID 58–SE-A1112.

18271 Muñoz, J.B., and Stern, C.R., 1989, Alkaline magmatism within the segment 38-39 S of the Plio-Quaternary volcanic belt of the southern South American continental margin: Journal of Geophysical Research, B v. 94, p. 4545–4560.

18430 Dimalanta, C.B., and Yumul, G.P., 2003, Magmatic and Amagmatic Contributions to Crustal Growth of an Island-Arc System: The Philippine Example: International Geology Review v. 45, p. 922–935.

18835 Colvin, A., Merrill, M., Demoor, M., Goss, A., and Varekamp, J.C., 2004, The Riscos Bayos Ignimbrites of the Caviahue-Copahue volcanic caldera complex, southern Andes, Argentina: EOS (Transactions, American Geophysical Union) v. 85, Joint Assembly Supplement, Abstract # V43A-16.

19497 Pappalardo, L., Piochi, M., DʼANTONIO, M., Civetta, L., and Petrini, R., 2002, Evidence for multi-stage magmatic evolution during the past 60 kyr at Campi Flegrei (Italy) deduced from Sr, Nd and Pb isotope data: Journal of Petrology v. 43, p. 1415–1434.

19753 Gertisser, R., Preece, K., and Keller, J., 2009, The Plinian Lower Pumice 2 eruption, Santorini, Greece: Magma evolution and volatile behaviour: Journal of Volcanology and Geothermal Research v. 186, p. 387–406.

19888 Reubi, O., 2004, Magmatic dynamics and volcanic evolution at Slamet volcano, Java and Batur volcanic field, Bali, Indonesia: PhD Dissertation, p. 215.

257

20700 Paoletti, V., Di, M., R., Cella, F., Florio, G., Motschka, K., Roberti, N., Secomandi, M., Supper, R., Fedi, M., and Rapolla, A., 2009, The Ischia volcanic island (Southern Italy): Inferences from potential field data interpretation: Journal of Volcanology and Geothermal Research v. 179, p. 69–86.

21022 Herve, F., Demant, A., Ramos, V.A., Pankhurst, R.J., and Suarez, M., 2000, The Southern Andes, in Cordani, U.G., Milani, E.J., and Filho, A.T., ed., Tectonic evolution of South America, p. 605–634.

21023 Arpa, M.C.B., 2005, Geochemistry of the upper Diliman Tuff unit in Manila, southwest Luzon, Philippines: Insights on its origin and comparison with Taal and Laguna caldera pyroclastic flows:

21291 Sruoga, P., Llambias, E.J., Fauquè, L., Schowandt, D., and Repol, D.G., 2005, Volcanological and geochemical evolution of the Diamante caldera-Maipo volcano complex in the southern Andes of Argentina (34o 10ʼS): Journal of South American Earth Sciences v. 19, p. 399–414.

21395 Takada, A., Yuasa, M., and Murakami, M., 1994, Geological maps of Aogashima Volcano and Submarine Volcanoes, south of Izu Islands: Chishitsu Chōsajo (Japan) no. 7 v. 1 sheet.

21789 Briggs, R.M., Gifford, M.G., Moyle, A.R., Normaln, M.D., Houghton, B.F., and Wilson, C.J.N., 1993, Geochemical zoning and eruptive mixing in ignimbrites from Mangakino volcano, Taupo Volcanic Zone, New Zealand: Journal of Volcanology and Geothermal Research v. 56, p. 175-203.

21957 Diraison, M., Cobbold, P.R., Gapais, D., Rossello, E.A., and Le, C., C., 2000, Cenozoic crustal thickening, wrenching and rifting in the foothills of the southernmost Andes: Tectonophysics v. 316, p. 91–119.

22203 Mohr, P., Mitchell, J.G., and Raynolds, R.G.H., 1980, Quaternary volcanism and faulting at OʼA caldera, central Ethiopian rift: Bulletin of Volcanology v. 43, p. 173– 189.

22218 Gavrilenko, M.G., Ozerov, A.Y., Kyle, P.R., and Eichelberger, J.C., 2006, Petrological and geochemical characteristics of magmatic melts at Gorely Volcano, Kamchatka, Russia: EOS (Transactions, American Geophysical Union) v. 87, Fall Meeting Supplement, Abstract VIIA-0558.

22380 Grunder, A.L., and Mahood, G.A., 1988, Physical and chemical models of zoned silicic magmas: the Loma Seca Tuff and Calabozos Caldera, Southern Andes: Journal of Petrology v. 29, p. 831–867.

22470 Kimura, J.I., Tanji, T., Yoshida, T., and Iizumi, S., 2001, Geology and geochemistry of lavas at Nekoma volcano: Implications for origin of Quaternary low-K andesite in the north-eastern Honshu arc, Japan: The Island Arc v. 10, p. 116–134.

22533 Sergeev, K.F., and Krasnyi, M.L., 1987, Geology-Geophysical Atlas of the Kuril- Kamchatka Island System: Leningrad, Ministerstvo Geologii SSSR.

22649 Katsui, Y., Yamagishi, H., Soya, T., Watanabe, Y., and Nakagawa, M., 1992, Cenozoic volcanism in southwestern Hokkaido, in Kato, H., and Noro, H., ed., 29th IGC Field Trip Guidebook, Vol. 4: Volcanoes and Geothermal Fields of Japan: Ibaraki, Japan, Geological Survey of Japan, p. 1–28.

258

22652 Ui, T., Kobayashi, T., and Suzuki-Kamata, K., 1992, Calderas and Pyroclastic Flows in Southern Kyushu, in Kato, H., and Noro, H., ed., 29th IGC Field Trip Guidebook, Vol. 4: Volcanoes and Geothermal Fields of Japan: Ibaraki, Japan, Geological Survey of Japan, p. 245–276.

22760 Yoshida, T., 2001, The evolution of arc magmatism in the NE Honshu arc, Japan: Tohoku Geophysical Journal v. 36, p. 131–149.

22831 Barberi, F., Santacroce, R., and Varet, J., 1974, Silicic peralkaline volcanic rocks of the Afar depression (Ethiopia): Bulletin of Volcanology v. 38, p. 755–790.

23212 Aramaki, S., 1984, Formation of the Aira Caldera, Southern Kyushu: Journal of Geophysical Research v. 89, p. 8485–8501.

23241 Ogawa, Y., _Nishida, Y., and Makino, M., 1994, A collision boundary imaged by magnetotellurics, Hidaka Mountains, central Hokkaido, Japan: Journal of Geophysical Research v. 99, p. 22–373.

23524 Yamada, E., 1988, Geologic development of the Onikobe caldera, Northeast Japan, with special reference to its hydrothermal system: Report of the Geological Survey of Japan v. 268, p. 61–190.

23587 Gonzalez-Ferran, O., 1997, Laguna del Maule Caldera: Southern Andean Range, Chile: IAVCEI General Assembly: Volcanic Activity and the Environment v. 14.

23805 Williams, H., and Meyer-Abich, H., 1955, Volcanism in the southern part of El Salvador, with particular reference to the collapse basins of lakes Coatepeque and Ilopango: University of California Publications in Geological Sciences v. 32, p. 1–64.

24053 Arakawa, Y., Kurosawa, M., Takahashi, K., Kobayashi, Y., Tsukui, M., and Amakawa, H., 1998, Sr-Ndisotopic and chemical characteristics of the silicic magma reservoir of the Aira pyroclastic eruption, southern Kyushu, Japan: Journal of Volcanology and Geothermal Research v. 80, p. 179–194.

24481 Varne, R., and Foden, J.D., 1986, Geochemical and isotopic systematics of eastern sunda arc volcanics: implications for mantle sources and mantle mixing processes: Origin of Arcs. Amsterdam: Elsevier v. 159–189.

25098 Donnelly, T.W., Beets, D., Carr, M.J., Jackson, T., Klaver, G., Lewis, J., Maury, R., Schellenkens, H., Smith, A.L., Wadge, G., and Westercamp, D., 1990, History and tectonic setting of Caribbean magmatism, The Caribbean region:: Boulder, p. 339–374.

25107 Ikeda, Y., Ikeda, T., and Kagami, H., 1990, Caldera-formation from geochemical aspects: A case study of the Toya caldera, southwestern Hokkaido, Japan: Journal of the Japanese Association of Mineralogy, Petrology, and Economic Geology v. 85, p. 569–577.

25224 Ramos, V.A., 2000, The Southern Central Andes, in Cordani, U.G., Milani, E.J., and Filho, A.T., ed., Tectonic evolution of South America: Rio de Janeiro, Brazil, In-Folo Producao Editorial, Grafica e Programacao Visual, p. 561–604.

25300 Miranda, F., Folguera, A., Leal, P.R., Naranjo, J.A., and Pesce, A., 2006, Upper Pliocene to Lower Pleistocene volcanic complexes and Upper Neogene deformation in the south-central Andes (36o 30ʼ-38oS), in Kay, S., and Ramos, V.,

259

ed., Evolution of an Andean Margin: Geol. Soc. Am. Spec. Paper 401: Boulder, CO, Geological Society of America, p. 287–298.

25720 De Silva, S.L., and Francis, P.W., 1991, Volcanoes of the Central Andes: Berlin, Springer-Verlag, 216 p.

25832 Bindeman, I.N., Ponomareva, V.V., Bailey, J.C., and Valley, J.W., 2004, Volcanic arc of Kamchatka: a province with high-18O magma sources and large-scale 18O/16O depletion of the upper crust: Geochimica Cosmochimica Acta v. 68, p. 841–865.

25927 Churikova, T., Dorendorf, F., and Worner, G., 2001, Sources and fluids in the mantle wedge below Kamchatka, evidence from across-arc geochemical variation: Journal of Petrology v. 42, p. 1567–1593.

25980 Piskunov, BN, Rybin, AV, and Sergeev, K.F., 1999, Petrogeochemical characteristics of rocks fom the Medvezhʼya caldera, the Iturup Island, Kuril Island: Doklady Earth Sciences v. 368, p. 989–992.

25981 Ban, M., Takahashi, K., Horie, T., and Toya, N., 2005, Petrogenesis of mafic inclusions in rhyolitic lavas from Narugo Volcano, Northeastern Japan: Journal of Petrology v. 46, p. 1543–1563.

26054 DePaolo, D.J., 1981, Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization: Earth and Planetary Science Letters v. 53, p. 189–202.

26511 Wolfe, J.A., and Self, S., 1983, Structural lineaments and Neogene volcanism in southwestern Luzon: The Tectonic and Geologic Evolution of Southeast Asian Seas and Islands v. 157–172.

26554 Aiuppa, A., Rotolo, S.G., and Villa, I.M., 1999, Stratigraphy, geochemistry and geochronology of a Quaternary pyroclastic sequence of the Chichontepec Volcano, El Salvador: Revista Geológica de América Central v. 22, p. 75–86.

26631 Hidalgo, P.J., Vogel, T.A., and Currier, R.M., 2006, Silicic ignimbrites of El Valle Volcano, Panama: Comparison with silicic ignimbrites of the Central American volcanic arc: Eos Transactions, American Geophysical Union v. 87, Abstract V33C–0695.

26793 Woodhead, J.D., Eggins, S.M., and Johnson, R.W., 1998, Magma Genesis in the New Britain Island Arc: Further Insights into Melting and Mass Transfer Processes: Journal of Petrology v. 39, p. 1641–1668.

27070 Higgins, M.W., 1973, Petrology of Newberry volcano, central Oregon: Geological Society of America Bulletin v. 84, p. 455.

27359 Rose, W.I.J., Hahn, G.A., Drexler, J.M., Malinconico, M.L., Peterson, P.S., and RL, W., 1981, Quaternary tephra of Northern Central America: Tephra Studies v. 193– 212.

27403 Survey, J.G., 2006, Quaternary Volcanoes of Japan Website:

27713 Hall, M.L., Samaniego, P., Le, P., JL, and Johnson, J.B., 2008, Ecuadorian Andes volcanism: A review of Late Pliocene to present activity: Journal of Volcanology

260

and Geothermal Research v. 176, p. 1–6.

27788 Iizasa, K., 1993, Assessment of the hydrothermal contribution to seafloor sediments in the Myojinsho submarine caldera, Shichito-Iwojima ridge, Izu- Ogasawara arc, Japan: Marine Geology v. 114, p. 119–132.

27901 Ferrari, L., Pasquare, G., Venegas-Salgado, S., and Romero-Rios, F., 2000, Geology of the western Mexican volcanic belt and adjacent Sierra Madre Occidental and Jalisco Block: Union Geofisica Mexicana symposium on Cenozoic tectonics and volcanism of Mexico, Puerto Vallarta, Mexico, 1994 v. 334, p. 65.

27935 Taylor, B., Klaus, A., Brown, G.R., Moore, G.F., Okamura, Y., and Murakami_, F., 1991, Structural development of Sumisu RIft, Izu-Bonin arc: Journal of Geophysical Research v. 96, p. 16113–16129.

27962 Hildreth, W., Lanphere, M.A., Champion, D.E., and Fierstein, J., 2004, Rhyodacites of Kulshan caldera, North Cascades of Washington: Postcaldera lavas that span the Jaramillo: Journal of Volcanology and Geothermal Research v. 130, p. 227– 264.

28082 Volynets, A., Woerner, G., Churikova, T., and Layer, P., 2006, The source of Pleistocene-Holocene volcanism in the Sredinny Range back arc, Kamchatka: constraints from trace elements and isotopic compositions, 5th Biennial Workshop on Subduction Processes emphasizing the Japan-Kurile-Kamchatka-Aleutian Arcs. Japan: p. 40–42.

28198 Tiffin, D.L., Taylor, B.D., Tufar, W., and Itikarai, I., 1990, A seabeam and sampling survey of newly discovered Tavui caldera near Rabaul, Papua New Guinea: SOPAC Cruise Report 132, p. 1-18.

28442 Ferrari, L., Lopez Martinez, M., Aguirre Diaz, G., and Carrasco Nunez, G., 1999, Space-time patterns of Cenozoic arc volcanism in central Mexico: From the Sierra Madre Occidental to the Mexican volcanic belt: Geology v. 27, p. 303–306.

28565 Fiske, R.S., Cashman, K.V., Shibata, A., and Watanabe, K., 2004, Tephra dispersal from Myojinsho Japan during its shallow submarine eruption of 1952- 1953: Bulletin of Volcanology v. 1953, p. 262–275.

28733 Innocenti, F., Kolios, N., Manetti, P., Rita, F., Villari, L., and Villari, L., 1982, Acid and basic late Neogene volcanism in central Aegean Sea: Its nature and geotectonic significance: Bulletin of Volcanology v. 45, p. 87–97.

29100 di Paola, G.M., 1971, Geology of the Corbetti Caldera area (Main Ethiopian Rift Valley): Bulletin of Volcanology v. 35, p. 497–506.

29220 Glasby, G.P., Iizasa, K., Yuasa, M., and Usui, A., 2000, Submarine hydrothermal mineralization on the Izu-Bonin Arc, South of Japan: An overview: Marine Georesources & Geotechnology v. 18, p. 141–176.

29580 Muñoz, J., and C, S., 1985, El complejo volcánico Pino Hachado en el sector nor- occidental de la Patagonia (38-39° S): Volcanismo pliocuaternario de trasarco en Sudamérica: Congreso Geològico Chileno v. 60, p. 380–412.

29891 Beauregard, J., 2001, Explosive rhyodacitic volcanism: the evolution and frequency of pre-1883 eruptions at Krakatau Volcano, Indonesia: PhD Dissertation, p. 358.

261

30411 Bardintzeff, J.M., and Deniel, C., 1992, Magmatic evolution of Pacaya and Cerro Chiquito volcanological complex, Guatemala: Bulletin of Volcanology v. 54, p. 267–283.

30573 Guest, J.E., Gaspar, J.L., Cole, P.D., Queiroz, G., Duncan, A.M., Wallenstein, N., Ferreira, T., and Pacheco, J.M., 1999, Volcanic geology of Furnas Volcano, Sao Miguel, Azores: Journal of Volcanology and Geothermal Research v. 92, p. 1–29.

30748 Gèze, B., Hudeley, H., Vincent, P., and Wacrenier, P., 1959, Les volcans du Tibesti (Sahara du Tchad): Bulletin of Volcanology v. 22, p. 135–172.

30889 McDermott, F., Delfin, J., FG, Defant, M.J., Turner, S., and Maury, R., 2005, The petrogenesis of volcanics from Mt. Bulusan and Mt. Mayon in the Bicol arc, the Philippines: Contributions to Mineralogy and Petrology v. 150, p. 652–670.

30958 Page, B.G.N., Bennett, J.D., Cameron, N.R., Bridge, D.M.C., Jeffery, D.H., Keats, W., Thaib, J., and Hepworth, J.V., 1979, A review of the main structural and magmatic features of northern Sumatra: Journal of the Geological Society of London v. 136, p. 569–579.

31852 Braitseva, O.A., Melekestsev, I.V., Ponomareva, V.V., and Sulerzhitsky, L.D., 1995, Ages of calderas, large explosive craters and active volcanoes in the Kuril- Kamchatka region, Russia: Bulletin of Volcanology v. 57, p. 383–402.

32101 Yamada, E., 1972, Study on the stratigraphy of Onikobe area, Miyagi Prefecture: Japan Geol Survey Bull v. 23, p. 217–231.

32815 Martinez, F., and Taylor, B., 1996, Backarc spreading, rifting, and microplate rotation, between transform faults in the Manus Basin: Marine Geophysical Researches v. 18, p. 203–224.

33696 Schmitt, A., De Silva, S., Trumbull, R., and Emmermann, R., 2001, Magma evolution in the Purico ignimbrite complex, northern Chile: evidence for zoning of a dacitic magma by injection of rhyolitic melts following mafic recharge: Contributions to Mineralogy and Petrology v. 140, p. 680–700.

34347 Kamata, H., 1989, Shishimuta caldera, the buried source of the Yabakei pyroclastic flow in the Hohi volcanic zone, Japan: Bulletin of Volcanology v. 51, p. 41–50.

34521 Rose, J., WI, Grant, N.K., and Easter, J., 1979, Geochemistry of the Los Chocoyos Ash, Quezaltenango Valley, Guatemala: Geological Society of America Special Paper v. 180, p. 87–99.

34719 Gardner, J.E., Burgisser, A., and Stelling, P., 2007, Eruption and Deposition of the Fisher Tuff (Alaska): Evidence for the Evolution of Pyroclastic Flows: The Journal of Geology v. 115, p. 417–435.

34845 Turner, S., Hawkesworth, C., Rogers, N., Bartlett, J., Worthington, T., Hergt, J., Pearce, J., and Smith, I., 1997, 238U/ 230Th disequilibria, magma petrogenesis, and flux rates beneath the depleted Tonga-Kermadec island arc: Geochimica et Cosmochimica Acta v. 61, p. 4855–4884.

34871 Ikeda, Y., 2002, Petrogenetic relationship between mafic magma of somma lavas and voluminous felsic magma with caldera formation: A case study of Akan volcanic field, eastern Hokkaido, Japan: Journal of Mineralogical and Petrological

262

Sciences v. 97, p. 39–46.

35291 Mandeville, C.W., Carey, S., and Sigurdsson, H., 1996, Magma mixing, fractional crystallization and volatile degassing during the 1883 eruption of Krakatau volcano, Indonesia: Journal of Volcanology and Geothermal Research v. 74, p. 243–274.

35338 Grib, E.N., and Leonov, V.L., 2004, The Evolution of Magma Chambers beneath Calderas in the Southern Sector of the Karymskii Volcanic Center: Part II. The PTF Conditions for the Crystallization of Ignimbrite-Forming Melts and the Evolution of the Magmatism, Vulkanologiya i Seysmologiya v. 5, p. 23–37.

35427 Morgan, L.E., and Renne, P.R., 2008, Diachronous dawn of Africaʼs Middle Stone Age: New 40Ar/39Ar ages from the Ethiopian Rift: Geology v. 36, p. 967.

35459 Melekestsev, I.V., Volynets, O.N., and Antonov, A.Y., 1997, Nemo III Caldera (Onekotan I., the northern Kuriles); structure, (super 14) C age, dynamics of the caldera-forming eruption, evolution of juvenile products: Volcanology and Seismology v. 19, p. 41–64.

35888 Bellon, H., Maury, R.C., Soeria-Atmadja, R., Cotten, J., and Polve, M., 2004, 65 my-long magmatic activity in Sumatra (Indonesia), from Paleocene to Present: Bulletin de la Societe Geologique de France v. 175, p. 61–72.

35902 Hall, M.L., and Mothes, P.A., 1997, Chacana caldera--The largest rhyolitic eruptive center in the Northern Andes: IAVCEI General Assembly v. 14.

36424 Garrison, J., Davidson, J., Reid, M., and Turner, S., 2006, Source versus differentiation controls on U-series disequilibria: Insights from Cotopaxi Volcano, Ecuador: Earth and Planetary Science Letters v. 244, p. 548–565.

36805 Hikaru, Y., Masahiro, Y., Satoshi, O., and Hirotsugu, N., 2003, Volcano- stratigraphy and geological development of Akaigawa Caldera from southwest Hokkaido, Japan: Japanese Magazine of Mineralogical and Petrological Sciences v. 32, p. 80–95.

38390 Ermakov, V.A., and Shteinberg, G.S., 1999, Kudryavyi Volcano and the evolution of Medvezhiya Caldera (Iturup Island, Kuril Islands.): Volcanology and Seismology v. 21, p. 307–338.

38514 Tormey, D.R., Hickey-Vargas, R., Frey, R.A., Lopez-Escobar, L., Harmon, R.S.E., and Rapela, C.W.E., 1991, Recent lavas from the Andean volcanic front (33-42° S): interpretations of along-arc compositional variations: Andean magmatism and its tectonic setting`: Andean magmatism and its tectonic setting, Geological Society of America Special Paper v. 265, p. 57–77.

38668 Ibrahim, A.K., Pontoise, B., Latham, G., and others., 1980, Structure of the New Hebrides arc-trench system: Journal of Geophysical Research, B v. 85, p. 253– 266.

39348 Folguera, A., and Ramos, V.A., 2000, Control estructural del volcàn Copahue (38S-71o): implicancias tectnicas para el arco volcànico cuaternario (36S-39S): Revista de la Asociaciòn Geològica Argentina v. 55, p. 229–244.

39606 Nakatsuka, Y., and Wada, K., Temporal changes of chemical composition of magmas in the last 10000 years at Mashu volcano, eastern Hokkaido, Japan:

263

39854 Machida, H., 2002, Volcanoes and tephras in the Japan area: Global Environmental Research- English Edition v. 6, p. 19–28.

40041 Beccaluva, L., Di, G., P., and Serri, G., 1991, Petrogenesis and tectonic setting of the Roman Volcanic Province, Italy: Lithos v. 26, p. 191–221.

40151 Barragan, R, Geist, D, Hall, M, Larson, P, and Kurz, M., 1998, Subduction controls on the compositions of lavas from the Ecuadorian Andes: Earth and Planetary Science Letters v. 154, p. 153–166.

40901 Priest, G.R., 1990, Volcanic and tectonic evolution of the Cascade volcanic arc, central Oregon: Journal of Geophysical Research, B v. 95, p. 19–583.

41341 Lindsay, J.M., Trumbull, R.B., and Siebel, W., 2005, Geochemistry and petrogenesis of late Pleistocene to Recent volcanism in Southern Dominica, Lesser Antilles: Journal of Volcanology and Geothermal Research v. 148, p. 253– 294.

41788 Kitamura, S., 2006, A preliminary report of the tephrochronological study of the eruptive history of Coatepeque Caldera, El Salvador, Central America: Bulletin of Faculty of Social Work, Hirosaki Gakuin University v. 8, p. 8–13.

41880 Geist, E.L., Childs, J.R., and Scholl, D.W., 1988, The origin of summit basins of the Aleutian ridge: implications for block rotation of an arc massif: Tectonics v. 7, p. 327–341.

42155 Defant, M.J., Jacques, D., Maury, R.C., de, B., J.Z., and Joron, J.L., 1989, Geochemistry and tectonic setting of the Luzon arc, Philippines: Bulletin of the Geological Society of America v. 101, p. 663–672.

42393 Mahood, G., and Hildreth, W., 1983, Nested calderas and trapdoor uplift at Pantelleria, Strait of Sicily: Geology v. 11, p. 722–726.

42728 Sussman, D., 1985, Apoyo caldera, Nicaragua: A major quaternary silicic eruptive center: Journal of Volcanology and Geothermal Research v. 24, p. 249–282.

42753 Spinks, K.D., Acocella, V., Cole, J.W., and Bassett, K.N., 2005, Structural control of volcanism and caldera development in the transtensional Taupo Volcanic Zone, New Zealand: Journal of Volcanology and Geothermal Research v. 144, p. 7-22.

43294 Alloway, B.V., Pribadi, A., and Westgate, J., 2004, Correspondence between glass-FT and 14 C ages of silicic pyroclastic flow deposits sourced from Maninjau caldera, west-central Sumatra: Earth and Planetary Science Letters v. 227, p. 121–133.

43947 Sakaguchi, K., and Yamada, E., 1988, The Kitagawa Dacite, pyroclastic flow deposits around the Onikobe caldera, northeast Japan: Report of the Geological Survey of Japan v. 268, p. 37–59.

44037 Goto, Y., Funayama, A., Gouchi, N., and Itaya, T., 2000, K-Ar ages of the Akan- Shiretoko volcanic chain lying oblique to the Kurile trench: Implications for tectonic control of volcanism: The Island Arc v. 9, p. 204–218.

44321 Ryburn, R.J., 1975, Talasea-Gasmata, New Britain-1: 250,000 geological series: Bureau of Mineral Resources, Geology & Geophysics, Australia, and Geological

264

Survey of Papua New Guinea, Explanatory Notes, SB/56-5 v. 9

44601 Yamada, E., 1988, Geologic development of the Onikobe caldera, Northeast Japan, with special reference to its hydrothermal system: Report Geological Survey of Japan v. 268, p. 61–190.

44673 Druitt, T.H., Brenchley, P.J., Gokten, Y.E., and Francaviglia, V., 1995, Late Quaternary rhyolitic eruptions from the Acigol Complex, central Turkey: Journal of the Geological Society v. 152, p. 655–667.

44696 Fitterman, D.V., 1988, Overview of the structure and geothermal potential of Newberry Volcano, Oregon: Journal of Geophysical Research, B v. 93, p. 10059– 10066.

44837 Nairn, I.A., Wood, C.P., and Bailey, R.A., 1994, The Reporoa Caldera, Taupo Volcanic Zone: source of the Kaingaroa Ignimbrites: Bulletin of Volcanology v. 56, p. 529–537.

45451 Druitt, T.H., Mellors, R.A., Pyle, D.M., and Sparks, R.S.J., 1989, Explosive volcanism on Santorini, Greece: Geological Magazine v. 126, p. 95–126.

45738 Fytikas, M., Innocenti, F., Manetti, P., Peccerillo, A., Mazzuoli, R., and Villari, L., 1984, Tertiary to Quaternary evolution of volcanism in the Aegean region: Geological Society London Special Publications v. 17, p. 687.

46265 Wunderman, R.L., and Rose, W.I., 1984, Amatitlan, an actively resurging cauldron 10 KM south of Guatemala City: Journal of Geophysical Research, B v. 89, p. 8525–8539.

46307 Yuasa, M., and Nohara, M., 1992, Petrographic and geochemical along-arc variations of volcanic rocks on the volcanic front of the Izu-Ogasawara (Bonin) arc: Bulletin of the Geological Survey of Japan v. 43, p. 421–456.

46455 Nakajima, J., and Hasegawa, A., 2003, Tomographic imaging of seismic velocity structure in and around the Onikobe volcanic area, northeastern Japan: implications for fluid distribution: Journal of Volcanology and Geothermal Research v. 127, p. 1–18.

46783 Lucassen, F., Becchio, R., Harmon, R., Kasemann, S., Franz, G., Trumbull, R., Wilke, H.G., Romer, R.L., and Dulski, P., 2001, Composition and density model of the continental crust at an active continental margin: The Central Andes between 21° and 27° S: Tectonophysics v. 341, p. 195–223.

46811 Erlich, E.N., and Volynets, O.N., 1979, Acid volcanism in Kamchatka: Bulletin of Volcanology v. 42, p. 175–254.

46942 Frey, F.A., Gerlach, D.C., Hickey, R.L., Lopez-Escobar, L., and Munizaga- Villavicencio, F., 1984, Petrogenesis of the Laguna del Maule volcanic complex, Chile (36° S): Contributions to Mineralogy and Petrology v. 88, p. 133–149.

47270 Peate, D.W., Pearce, J.A., Hawkesworth, C., Colley, H., Edwards, C.M.H., and Hirose, K., 1997, Geochemical variations in Vanuatu Arc lavas: the role of subducted material and a variable mantle wedge tomposition: Journal of Petrology v. 38, p. 1331–1358.

265

47380 Sigvaldason, G.E., 2002, Volcanic and tectonic processes coinciding with glaciation and crustal rebound: an early Holocene rhyolitic eruption in the Dyngjufjoll volcanic centre and the formation of the Askja caldera, north Iceland: Bulletin of Volcanology v. 64, p. 192–205.

47529 Kondo, H., Tanaka, K., Mizuochi, Y., and Ninomiya, A., 2004, Long-term changes in distribution and chemistry of middle Miocene to Quaternary volcanism in the Chokai-Kurikoma area across the Northeast Japan Arc: The Island Arc v. 13, p. 18.

47561 Zouzias, D., and St, S., K., 2008, Consanguineous geochemistry of the Kos Plateau and Tilos D and E Pumices, Aegean Volcanic Arc, Hellas: Neues Jahrbuch Mineralogie-Abhandlungen v. 184, p. 231–241.

47739 Ramos, V.A., and Kay, S.M., 2006, Overview of the tectonic evolution of the southern Central Andes of Mendoza and Neuquén (35°--39° S latitude): GSA Special Papers v. 407, p. 1–18.

48154 Stern, C.R., 2004, Active Andean volcanism: Its geologic and tectonic setting: Revista Geolgica de Chile v. 31, p. 161–206.

48205 Van Avendonk, H.J.A., Shillington, D.J., Holbrook, W.S., and Hornbach, M.J., 2004, Inferring crustal structure in the Aleutian island arc from a sparse wide- angle seismic data set: Geochemistry Geophysics Geosystems v. 5, p. doi:10.1029/2003GC000664.

48209 Deering, C.D., Vogel, T.A., Patino, L.C., and Alvarado, G.E., 2005, Petrogenesis of distinct silicic magma types from the lower Pleistocene Guachipelin Caldera, NW Costa Rica: extensive magma mixing and protracted subvolcanic residence: Abstracts with Programs - Geological Society of America v. 37, p. 308.

48225 Smith, V.C., Shane, P., and Nairn, I.A., 2005, Trends in rhyolite geochemistry, mineralogy, and magma storage during the last 50 kyr at Okataina and Taupo volcanic centres, Taupo Volcanic Zone, New Zealand: Journal of Volcanology and Geothermal Research v. 148, p. 372–406.

48347 Fiske, R.S., Naka, J., Iizasa, K., Yuasa, M., and Klaus, A., 2001, Submarine silicic caldera at the front of the Izu-Bonin arc, Japan: Voluminous seafloor eruptions of rhyolite pumice: Bulletin of the Geological Society of America v. 113, p. 813–824.

48583 Hildreth, W., Fierstein, J., and Lanphere, M., 2003, Eruptive history and geochronology of the Mount Baker volcanic field, Washington: Bulletin of the Geological Society of America v. 115, p. 729–764.

48650 Hall, M.L., Mothes, P.A., and Eissen, J.-P., 2000, Rhyolitic magma body and ascending basic andesites: Bimodal Cotopaxi magmatism: EOS (Transactions, American Geophysical Union) v. 81, p. 1309.

49119 Hallinan, S., 1993, Nonchaotic collapse at funnel calderas; gravity study of the ring fractures at Guayabo Caldera, Costa Rica: Geology v. 21, p. 367–370.

49716 Hunter, A.G., and Blake, S., 1995, Petrogenetic evolution of a transitional tholeiitic- -calc-alkaline series: Towada Volcano, Japan: Journal of Petrology v. 36, p. 1579–1605.

266

49938 Greene, H.G., Macfarlane, A., Johnson, D.P., and Crawford, A.J., 1988, Structure and tectonics of the central New Hebrides Arc, in Greene, H.G., and Wong, F.L., ed., Geology and offshore resources of Pacific island arcs—Vanuatu region: Houston, TX, Circum-Pacific Council for Energy and Mineral Resources,

50068 Hildreth, H., Grunder, A.L., and Drake, R.E., 1984, The Loma Seca Tuff and the Calabozos Caldera; a major ash-flow and caldera complex in the Southern Andes of central Chile: Bulletin of the Geological Society of America v. 95, p. 45–54.

50421 Melekestsev, I.V., Felitsyn, S.B., and Kirʼyanov, V.Y., 1991, Izverzheniye vulkana Opala okolo 500 g.; krupneysheye eksplosivnoye izverzheniye nashey ery na Kamchatke. Eruption of the Opala Volcano in 500 AD; the largest explosive eruption in Kamchatka in the new era: Vulkanologiya i Seysmologiya v. 1, p. 21– 34.

50459 Reubi, O., and Nicholls, I.A., 2005, Structure and dynamics of a silicic magmatic system associated with caldera-forming eruptions at Batur Volcanic Field, Bali, Indonesia: Journal of Petrology v. 46, p. 1367–1391.

51068 Dunbar, N.W., and Hervig, R.L., Volatile and trace element composition of melt inclusions from the Lower Bandelier Tuff: Implications for magma chamber processes and eruptive style: Journal of Geophysical Research, B v. 97

51071 Harmon, R.S., 2005, The Geological Development of Panama: Water Science and Technology Library v. 52, p. 45.

51494 Alvarado, G.E., Dengo, C., Martens, U., Bundschuh, J., Aguilar, T., and Bonis, S.B., 2007, Chapter 13: Stratigraphy and geologic history, in Bundschuh, J., and Alvarado, G.E., ed., Central America: Geology, Resources, Hazards vol. 1: London, UK, Taylor and Francis, p. 345–394.

51846 Hall, M., and Mothes, P., 2008, The rhyolitic-andesitic eruptive history of Cotopaxi volcano, Ecuador: Bulletin of Volcanology v. 70, p. 675–702.

52414 Sakamoto, I., and Tanimoto, H., 1996, Rhyolitic pumice dredged from the bank located between Hachijojima and Aogashima; the morphological features of subaqueous erupted pumice: Journal of the Faculty of Marine Science and Technology, Tokai University v. 41, p. 139–156.

53610 Sigurdsson, H., 1972, Partly-welded pyroclast flow deposits in Dominica, Lesser Antilles: Bulletin of Volcanology v. 36, p. 148–163.

53643 McClusky, S., Balassanian, S., Barka, A., Demir, C., Ergintav, S., Georgiev, I., Gurkan, O., Hamburger, M., Hurst, K., Kahle, H., and others, 2000, Global Positioning System constraints on plate kinematics and dynamics in the eastern Mediterranean and Caucasus: Journal of Geophysical Research v. 105

54794 Goss, A., Kay, S.M., and Mpodozis, C., 2009, The chemistry of a dying continental arc: The Incapillo caldera and dome complex (~28o S): Contributions to Mineralogy and Petrology, in review

55571 Nakamura, Y., Yokose, H., Matsui, T., Yokose, R., Yasuhara, M., Yokose, H., Knesel, K.M., Duffield, W.A., Priest, S.S., Yokose, H., and others, 1999, Petrogenesis of large scale felsic magmas from Kikai, Ata and Aira caldera, southern Kyushu, Japan: EOS Transactions, American Geophysical Union v. V21A-02

267

56090 Allen, S.R., 2001, Reconstruction of a major caldera-forming eruption from pyroclastic deposit characteristics: Kos Plateau Tuff, eastern Aegean Sea: Journal of Volcanology and Geothermal Research v. 105, p. 141-162.

56985 Orr, W., and Orr, E., 1996, Geology of the Pacific Northwest, 2nd ed.: Boston, McGraw Hill,

58071 Brown, S.J.A., Wilson, C.J.N., Cole, J.W., and Wooden, J., 1998, The Whakamaru group ignimbrites, Taupo Volcanic Zone, New Zealand: evidence for reverse tapping of a zoned silicic magmatic system: Journal of Volcanology and Geothermal Research v. 84, p. 1–37.

58200 Mechie, J., Keller, G.R., Prodehl, C., Gaciri, S., Braile, L.W., Mooney, W.D., Gajewski, D., and Sandmeier, K.J., 1994, Crustal structure beneath the Kenya Rift from axial profile data: Tectonophysics v. 236, p. 179–200.

58433 Turner, S., and Hawkesworth, C., 1997, Constraints on flux rates and mantle dynamics beneath island arcs from Tonga-Kermadec lava geochemistry: Nature v. 389, p. 568–573.

58698 Camus, G., Gourgaud, A., and Vincent, P.M., 1987, Petrologic evolution of Krakatau(Indonesia): implications for a future activity: Journal of Volcanology and Geothermal Research v. 33, p. 299–316.

61123 Goss, A.R., Kay, S.M., Mpodozis, C., and Singer, B.S., 2009, The Incapillo Caldera and Dome Complex (~ 28° S, Central Andes): A stranded magma chamber over a dying arc: Journal of Volcanology and Geothermal Research v. 184, p. 389–404.

61575 Whitford, D.J., 1975, Strontium isotope studies of the volcanic rocks of the Sunda Arc, Indonesia, and their petrogenetic significance: Geochimica et Cosmochimica Acta v. 39, p. 1287–1302.

61960 Tolstykh, M.L., Naumov, V.B., Babansky, A.D., Bogoyavlenskaya, G.E., and Khubunaya, S.A., 2003, Chemical Composition, Volatile Components, and Trace Elements in Andesitic Magmas of the Kurile-Kamchatka Region: Petrology v. 11, p. 407–425.

62000 Ikeda, Y., Ikeda, N., and Kagami, H., 1990, Caldera-formation from geochemical aspects: a case study of the Toya caldera, southwestern Hokkaido, Japan: Journal of Mineralogy Petrology and Economic Geology v. 85, p. 569–577.

62805 Liu, R., Fan, Q., Zheng, X., Zhang, M., and Li, N., 1998, The magma evolution of Tianchi volcano, Changbaishan: Science in China Series D: Earth Sciences v. 41, p. 382–389.

63192 Permenter, J.L., and Oppenheimer, C., 2007, Volcanoes of the Tibesti massif (Chad, northern Africa): Bulletin of Volcanology v. 69, p. 609–626.

63677 Foden, J.D., 1983, The petrology of the calcalkaline lavas of Rindjani volcano, East Sunda arc: a model for island arc petrogenesis: Journal of Petrology v. 24, p. 98.

63891 Japan, G.S.o., 2009, Seamless digital geological map of Japan 1: 200,000. Research Information Database DB084, Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology.

268

63969 Lacasse, C., Sigurdsson, H., Carey, S.N., Johannesson, H., Thomas, L.E., and Rogers, N.W., 2007, Bimodal volcanism at the Katla subglacial caldera, Iceland: insight into the geochemistry and petrogenesis of rhyolitic magmas: Bulletin of Volcanology v. 69, p. 373–399.

64410 Hall, M.L., and Mothes, P.A., 2008, The Chacana caldera complex in Ecuador: IOP Conference Series: Earth and Environmental Science v. 13, p. 012004.

65852 Mahood, G.A., 1981, Chemical evolution of a Pleistocene rhyolitic center: Sierra la Primavera, Jalisco, Mexico: Contributions to Mineralogy and Petrology v. 77, p. 129–149.

65957 Civetta, L., Gallo, G., and Orsi, G., 1991, Sr-and Nd-isotope and trace element constraints on the chemical evolution of the magmatic system of Ischia (Italy) in the last 55 ka: Journal of Volcanology and Geothermal Research v. 46, p. 213– 230.

67024 Petrinovic, I.A., Riller, U., and Brod, J.A., 2005, The Negra Muerta Volcanic Complex, southern Central Andes: geochemical characteristics and magmatic evolution of an episodically active volcanic centre: Journal of Volcanology and Geothermal Research v. 140, p. 295–320.

67103 Stix, J., and Layne, G.D., 1996, Gas saturation and evolution of volatile and light lithophile elements in the Bandelier magma chamber between two caldera- forming eruptions: Journal of Geophysical Research v. 101, p. 25–25.

67315 Wilson, C.J.N., Houghton, B.F., McWilliams, M.O., Lanphere, M.A., Weaver, S.D., and Briggs, R.M., 1995, Volcanic and structural evolution of Taupo Volcanic Zone, New Zealand: a review: Journal of Volcanology and Geothermal Research v. 68, p. 1–28.

68284 Steck, L.K., Thurber, C.H., Fehler, M.C., Lutter, W.J., Roberts, P.M., Baldridge, W.S., Stafford, D.G., and Sessions, R., 1998, Crust and upper mantle P wave velocity structure beneath Valles caldera, New Mexico: Results from the Jemez teleseismic tomography experiment: Journal of Geophysical Research, B v. 103

68362 Kabeto, K., Sawada, Y., and Roser, B., 2009, Compositional Differences between Felsic Volcanic rocks from the Margin and Center of the Northern Main Ethiopian Rift: Mamona Ethiopian Journal of Science v. 1

68526 Nakamura, Y., Hirakawa, K., and Naganuma, T., 1999, Tephras at and around a Paleolithic Site in the Shirataki Basin, Eastern Hokkaido, Japan: Journal of Geography - Tokyo v. 108, p. 616–628.

69763 Finlayson, D.M., Gudmundsson, O., Itikarai, I., Nishimura, Y., and Shimamura, H., 2003, Rabaul volcano, Papua New Guinea: seismic tomographic imaging of an active caldera: Journal of Volcanology and Geothermal Research v. 124, p. 153– 171.

69857 Agostini, S., Corti, G., Doglioni, C., Carminati, E., Innocenti, F., Tonarini, S., Manetti, P., Di, V., G., and Montanari, D., 2006, Tectonic and magmatic evolution of the active volcanic front in El Salvador: insight into the Berlín and Ahuachapán geothermal areas: Geothermics v. 35, p. 368–408.

69865 Eguez, A., Alvarado, A., Yepes, H., Machette, M.N., Costa, C., and Dart, R.L., 2003, Database and map of Quaternary faults and folds of Ecuador and its

269

offshore regions: USGS Open-File Report 03–289.

70562 Woolley, A.R., and Kogarko, L.N., 2001, Alkaline Rocks and Carbonatites of the World: Africa: Geological Society Pub House,

71508 John, D.A., Henry, C.D., and Colgan, J.P., 2008, Magmatic and tectonic evolution of the Caetano caldera, north-central Nevada: A tilted, mid-Tertiary eruptive center and source of the Caetano Tuff: Geosphere v. 4, p. 75.

71846 Le, T., C., Tiercelin, J.J., Gibert, E., Travi, Y., Lezzar, K.E., Richert, J.P., Massault, M., Gasse, F., Bonnefille, R., Decobert, M., and others., 1999, The Ziway-Shala lake basin system, Main Ethiopian Rift: influence of volcanism, tectonics, and climatic forcing on basin formation and sedimentation: Palaeogeography Palaeoclimatology Palaeoecology v. 150, p. 135–178.

72191 Wolff, J.A., Ramos, F.C., and Davidson, J.P., 1999, Sr isotope disequilibrium during differentiation of the Bandelier Tuff: constraints on the crystallization of a large rhyolitic magma chamber: Geology v. 27, p. 495.

72245 Kamata, H., Danhara, T., Yamashita, T., Hoshizumi, H., and Hayashida, A., 1994, Correlation of the Azuki volcanic ash of the Osaka Group and the Ku6C volcanic ash of the Kazusa Group to the Imaichi pyroclastic-flow deposit in central Kyushu, Japan─ A co-ignimbrite ash erupted from Shishimuta: Journal of the Geologic Society of Japan v. 100, p. 848–866.

72444 Rubio, E., 2000, Crustal structure of the southernmost Chilean margin from seismic and gravity data: Tectonophysics v. 323, p. 39–60.

72807 Iizasa, K., Fiske, R.S., Ishizuka, O., Yuasa, M., Hashimoto, J., Ishibashi, J., Naka, J., Horii, Y., Fujiwara, Y., Imai, A., and others, 1999, A Kuroko-type polymetallic sulfide deposit in a submarine silicic caldera: Science v. 283, p. 975–977.

72954 Hammersley, L., and DePaolo, D.J., 2002, Oxygen isotope evidence for the role of crustal contamination in the evolution of the Chalupas caldera system, Northern Andes, Ecuador: EOS Transactions, American Geophysical Union v. 83, p. 47.

73618 Zhuravlev, D., Tsvetkov, A.A., Zhuravlev, A.Z., Gladkov, N.G., and Chernyshev, I.V., 1987, 143Nd/144Nd and 87Sr/86Sr ratios in recent magmatic rocks of the Kurile Island Arc: Chemical Geology v. 66, p. 227–243.

74300 Nairn, I.A., Mckee, C.O., Talai, B., and Wood, C.P., 1995, Geology and eruptive history of the Rabaul Caldera area, Papua New Guinea: Journal of Volcanology and Geothermal Research v. 69, p. 255–284.

74913 Milner, D.M., Cole, J.W., and Wood, C.P., 2003, Mamaku Ignimbrite: a caldera- forming ignimbrite erupted from a compositionally zoned magma chamber in Taupo Volcanic Zone, New Zealand: Journal of Volcanology and Geothermal Research v. 122, p. 243-264.

75089 Ewart, A., and Stipp, J.J., 1968, Petrogenesis of the volcanic rocks of the central North Island, New Zealand, as indicated by a study of Sr 87/Sr 86 ratios, and Sr, Rb, K, U and Th abundances: Geochimica et Cosmochimica Acta v. 32, p. 699– 736.

75292 Defant, M.J., Maury, R.C., Ripley, E.M., Feigenson, M.D., and Jacques, D., 1991, An Example of Island-Arc Petrogenesis: Geochemistry and Petrology of the

270

Southern Luzon Arc, Philippines: Journal of Petrology v. 32, p. 455–500.

75376 Beresford, S.W., Cole, J.W., and Weaver, S.D., 2000, Weak chemical and mineralogical zonation in the Kaingaroa Ignimbrite, Taupo Volcanic Zone, New Zealand: New Zealand Journal of Geology and Geophysics v. 43, p. 639–650.

75495 Chigira, M., Nakamoto, M., and Nakata, E., 2002, Weathering mechanisms and their effects on the landsliding of ignimbrite subject to vapor-phase crystallization in the Shirakawa pyroclastic flow, northern Japan: Engineering Geology v. 66, p. 111–125.

77113 Li, X., Bock, G., Vafidis, A., Kind, R., Harjes, H.P., Hanka, W., Wylegalla, K., Van Der Meijde, M., and Yuan, X., 2003, Receiver function study of the Hellenic subduction zone: imaging crustal thickness variations and the oceanic Moho of the descending African lithosphere: Geophysical Journal International v. 155, p. 733-748.

78879 Rangin, C., Le, P., X., Mazzotti, S., Pubellier, M., Chamot-Rooke, N., Aurelio, M., Walpersdorf, A., and Quebral, R., 1999, Plate convergence measured by GPS across the Sundaland/Philippine Sea Plate deformed boundary: the Philippines and eastern Indonesia: Geophysical Journal International v. 139, p. 296–316.

78901 Blake, D.H., and Ewart, A., 1974, Petrography and geochemistry of the Cape Hoskins volcanoes, New Britain, Papua New Guinea: Australian Journal of Earth Sciences v. 21, p. 319–331.

79342 Kamata, H., and Kodama, K., 1994, Tectonics of an arc-arc junction: an example from Kyushu Island at the junction of the Southwest Japan arc and the Ryukyu arc: Tectonophysics v. 233, p. 69–81.

80193 Leat, P.T., 1984, Geological evolution of the trachytic caldera volcano Menengai, Kenya Rift valley: Journal of the Geological Society v. 141, p. 1057-1069.

80210 Pasyanos, M.E., and Nyblade, A.A., 2007, A top to bottom lithospheric study of Africa and Arabia: Tectonophysics v. 444, p. 27–44.

80667 Baranov, B.V., Werner, R., Hoernle, K.A., Tsoy, I.B., van, d.B., P., and Tararin, I.A., 2002, Evidence for compressionally induced high subsidence rates in the Kurile Basin (Okhotsk Sea): Tectonophysics v. 350, p. 63–97.

80822 Worthington, T.J., Gregory, M.R., and Bondarenko, V., 1999, The Denham Caldera on Raoul Volcano: dacitic volcanism in the Tonga--Kermadec arc: Journal of Volcanology and Geothermal Research v. 90, p. 29–48.

80895 Dreher, S.T., Eichelberger, J.C., and Larsen, J.F., 2005, The Petrology and Geochemistry of the Aniakchak Caldera-forming Ignimbrite, Aleutian Arc, Alaska: Journal of Petrology v. 46, p. 1747–1768.

82126 Naumov, V.B., Tolstykh, M.L., Grib, Y.N., Leonov, V.L., and Kononkova, N.N., 2008, Chemical composition, volatiles, and trace elements in melts of Karymskiy volcanic center (Kamchatka) and Golovnina Volcano (Kunashir Island) from data from inclusions in minerals: Petrologiya v. 16, p. 3–20.

83033 Druitt, T.H., Edwards, L., Mellors, R.M., Pyle, D.M., Sparks, R.S.J., Lanphere, M., Davies, M., and Barriero, B., 1999, Geochemistry, isotopic composition and petrogenetic modelling of the Thera pyroclastics, Santorini: Geological Society,

271

London, p. 89–116.

83109 Barberi, F., Ghigliotti, M., Macedonio, G., Orellana, H., Pareschi, M.T., and Rosi, M., 1992, Volcanic hazard assessment of Guagua Pichincha (Ecuador) based on past behaviour and numerical models: Journal of Volcanology and Geothermal Research v. 49, p. 53–68.

83513 Barberi, F., Marinelli, G., Santacrosse, R., Tazieff, H., Varet, J., Faure, H., and Giglia, G., 1973, Geology of northern Afar (Ethiopia): Revue de Geographie Physique et de Geolgie Dynamique v. 15, p. 433–490.

83878 Nakamura, Y., Yokose, H., Yasuhara, M., Yokose, H., Soriano, C., Martì, J., Dingwell, D.B., Dawson, P.B., Chouet, B.A., Okubo, P., and others, 1998, Origin of Rhyodacite Magma from Ata Caldera South Kyushu, Japan: Implication: American Geophysical Union 1998 Fall meeting, Abstract V22A-05

83893 Bursik, M., 2009, A general model for tectonic control of magmatism: Examples from Long Valley Caldera (USA) and El Chichòn (México): Geofìsica Internacional v. 48, p. 171–183.

83940 Lagmay, A.M.F., Tengonciang, A.M.P., and Uy, H.S., 2005, Structural setting of the Bicol Basin and kinematic analysis of fractures on Mayon Volcano, Philippines: Journal of Volcanology and Geothermal Research v. 144, p. 23–36.

84468 Banner, J.L., Bence, A.E., and Johnson, R., 1980, Petrology of an Island Arc Volcanic Complex: EOS Transactions, American Geophysical Union v. 61, p. 401.

84647 Lara, L.E., and Folguera, A., 2006, The Pliocene to Quaternary narrowing of the Southern Andean volcanic arc between 37° and 41°S latitude, in Kay, S.M., and Ramos, V., ed., Evolution of an Andean Margin, Geol. Soc. Am. Spec. Paper 401: Boulder, CO, Geological Society of America, p. 299–315.

85085 Grabkov, V.K., 1973, New data on the geologic structure of Iturup Island: Doklady Earth Sciences v. 1973, p. 104.

85372 Leonov, V.L., and Griben, E.N., 1998, Kalʼdery i ignimbrites Uzon-Semyachinskogo rayona, Kamchatka; novyye dannyye po rezulʼtatam razrezov plato Shirokoye. Calderas and ignimbrites on the Uzon-Semyachinsk region, Kamchatka; new evidence from a study of section in the Shirokoye Plateau: Vulkanologiya i Seysmologiya v. 1998, p. 41–59.

85459 Trenkamp, R., Kellogg, J.N., Freymueller, J.T., and Mora, H.P., 2002, Wide plate margin deformation, southern Central America and northwestern South America, CASA GPS observations: Journal of South American Earth Sciences v. 15, p. 157–171.

85901 Coats, R.R., 1959, Geologic reconnaissance of Semisopochnoi Island, Western Aleutian Islands, Alaska: US Geol: U.S. Geological Survey Bulletin Report: B 1028-O, p. 477–519.

87328 Chan, L.H., Leeman, W.P., and You, C.F., 1999, Lithium isotopic composition of Central American Volcanic Arc lavas: implications for modification of subarc mantle by slab-derived fluids: Chemical Geology v. 160, p. 255–280.

87427 Izbekov, P., Gardner, J.E., and Eichelberger, J.C., 2004, Comagmatic granophyre and dacite from Karymsky volcanic center, Kamchatka: experimental constraints for magma storage conditions: Journal of Volcanology and Geothermal Research

272

v. 131, p. 1-18.

87667 Sutherland, R., 1999, Basement geology and tectonic development of the greater New Zealand region: an interpretation from regional magnetic data: Tectonophysics v. 308, p. 341–362.

88259 Ozawa, A., Tagami, T., Listanco, E.L., Arpa, C.B., and Sudo, M., 2004, Initiation and propagation of subduction along the Philippine Trench: evidence from the temporal and spatial distribution of volcanoes: Journal of Asian Earth Sciences v. 23, p. 105–111.

88703 McGarvie, D.W., Burgess, R., Tindle, A.G., Tuffen, H., and Stevenson, J.A., 2006, Pleistocene rhyolitic volcanism at the Torkfajökull central volcano, Iceland: eruption ages, ice-volcano interactions, and geochemical evolution: Jökull v. 56, p. 57–75.

88814 Kamata, H., 1989, Volcanic and structural history of the Hohi volcanic zone, central Kyushu, Japan: Bulletin of Volcanology v. 51, p. 315–332.

88869 Civetta, L., DʼAntonio, M., Orsi, G., and Tilton, G.R., 1998, The geochemistry of volcanic rocks from Pantelleria Island, Sicily Channel: petrogenesis and characteristics of the mantle source region: Journal of Petrology v. 39, p. 1453.

89053 Masuda, Y., Nishimura, S., Ikeda, T., and Katsui, Y., 1975, Rare-earth and trace elements in the Quaternary volcanic rocks of Hokkaido, Japan: Chemical Geology v. 15, p. 251–271.

89324 Wei, H., Wang, Y., Jin, J., Gao, L., Yun, S.H., and Jin, B., 2007, Timescale and evolution of the intracontinental Tianchi volcanic shield and ignimbrite-forming eruption, Changbaishan, Northeast China: Lithos v. 96, p. 315–324.

89349 Katsui, Y., 1961, Petrochemistry of the Quaternary volcanic rocks of Hokkaido and surrounding areas: Journal of Faculty of Science, Hokkaido University v. 11, p. 1– 58.

89524 Hildyard, S.C., Cole, J.W., and Weaver, S.D., 2000, Tikorangi Ignimbrite: a 0.89 Ma mixed andesite-rhyolite ignimbrite, Matahana Basin, Taupo Volcanic Zone, New Zealand: New Zealand Journal of Geology and Geophysics v. 43, p. 95-107.

89687 Harry, D.L., and Green, N.L., 1999, Slab dehydration and basalt petrogenesis in subduction systems involving very young oceanic lithosphere: Chemical Geology v. 160, p. 309–333.

89755 Halliday, A.N., Fallick, A.E., Hutchinson, J., and Hildreth, W., 1984, Nd, Sr and O isotopic investigation into the causes of chemical and isotopic zonation in the Bishop Tuff, California: Earth and Planetary Science Letters v. 68, p. 379-391.

89829 Hildreth, W., and Moorbath, S., 1988, Crustal contributions to arc magmatism in the Andes of Central Chile: Contributions to Mineralogy and Petrology v. 98, p. 455–489.

89974 Bailey, R.A., and Carr, R.G., 1994, Physical geology and eruptive history of the Matahina Ignimbrite, Taupo Volcanic Zone, North Island, New Zealand: New Zealand Journal of Geology and Geophysics v. 37, p. 319–344.

273

90083 Esperança, S., and Crisci, G.M., 1995, The island of Pantelleria: a case for the development of DMM-HIMU isotopic compositions in a long-lived extensional setting: Earth and Planetary Science Letters v. 136, p. 167–182.

90443 Gravley, D.M., Wilson, C.J.N., Rosenberg, M.D., and Leonard, G.S., 2006, The nature and age of Ohakuri Formation and Ohakuri Group rocks in surface exposures and geothermal drillhole sequences in the central Taupo Volcanic Zone, New Zealand: New Zealand Journal of Geology and Geophysics v. 49, p. 305–308.

91273 Bacon, C.R., Gunn, S.H., Lanphere, M.A., and Wooden, J.L., 1994, Multiple isotopic components in Quaternary volcanic rocks of the Cascade arc near Crater Lake, Oregon: Journal of Petrology v. 35, p. 1521–1556.

91515 Özdemir, Y., Karaoğlu, Ö., Tolluoğlu, A., and Güleç, N., 2006, Volcanostratigraphy and petrogenesis of the Nemrut stratovolcano (East Anatolian High Plateau): The most recent post-collisional volcanism in Turkey: Chemical Geology v. 226, p. 189–211.

91608 Gertisser, R., and Keller, J., 2003, Trace element and Sr, Nd, Pb and O isotope variations in medium-K and high-K volcanic rocks from Merapi Volcano, Central Java, Indonesia: Evidence for the involvement of subducted sediments in Sunda Arc magma genesis: Journal of Petrology v. 44, p. 457–489.

93681 Verma, S.P., 2001, Geochemical and Sr--Nd--Pb isotopic evidence for a combined assimilation and fractional crystallisation process for volcanic rocks from the Huichapan caldera, Hidalgo, Mexico: Lithos v. 56, p. 141–164.

93712 Calderone, G.M., Gronvold, K., and Oskarsson, N., 1990, The welded air-fall tuff layer at Krafla, Northern Iceland--a composite eruption triggered by injection of basaltic magma: Journal of Volcanology and Geothermal Research v. 44, p. 303– 314.

93981 Fiske, R.S., Jiro, N., IIzasa, K., and Yuasa, M., 1995, Calderaforming-Submarine Pyroclastic Erup-tion at Myojin Knoll, IzuBonin Are: JAMSTEC Journal of Deep Sea Research v. 11, p. 315–322.

94049 Hunter, A.G., 1998, Intracrustal Controls on the Coexistence of Tholeiitic and Calc- alkaline Magma Series at Aso Volcano, SW Japan: Journal of Petrology v. 39, p. 1255–1284.

94195 Yokoyama, H., Yahata, M., Okamura, S., and Nishido, H., 2003, Volcano- stratigraphy and geological development of Akaigawa Caldera from southwest Hokkaido, Japan: Japanese Magazine of Mineralogical and Petrological Sciences v. 32, p. 80–95.

94225 Mohr, P., 1978, Afar: Annual Review of Earth and Planetary Sciences v. 6, p. 145– 172.

94372 Wilson, C.J., Blake, S., Charlier, B.L., and Sutton, A., 2006, The 26.5 ka Oruanui Eruption, Taupo Volcano, New Zealand: Development, Characteristics and Evacuation of a Large Rhyolitic Magma Body: Journal of Petrology v. 47, p. 35– 69.

94818 Hildreth, W., Drake, R.E., Godoy, E., and Munizaga, F., 1991, Bobadilla caldera and 1.1 Ma ignimbrite at Laguna del Maule, Central Chile: Congreso Geologico

274

Chileno v. 6, p. 62–63.

94832 Chesner, C., 1998, Petrogenesis of the Toba Tuffs, Sumatra, Indonesia: Journal of Petrology v. 39, p. 397–438.

94922 Geyer, A., and Martì, J., 2008, The new worldwide collapse caldera database (CCDB): A tool for studying and understanding caldera processes: Journal of Volcanology and Geothermal Research v. 175, p. 334–354.

96829 Skilling, I.P., 1993, Incremental caldera collapse of Suswa volcano, Gregory Rift Valley, Kenya: Journal of the Geological Society v. 150, p. 885–896.

96840 Hasegawa, A., Nakajima, J., Umino, N., and Miura, S., 2005, Deep structure of the northeastern Japan arc and its implications for crustal deformation and shallow seismic activity: Tectonophysics v. 403, p. 59–75.

96977 Scott, S.C., and Skilling, I.P., 1999, The role of tephrachronology in recognizing synchronous caldera-forming events at the Quaternary volcanoes Longonot and Suswa, south Kenya Rift: Geological Society London Special Publications v. 161, p. 47.

97525 Druitt, T.H., and Francaviglia, V., 1992, Caldera formation on Santorini and the physiography of the islands in the late Bronze Age: Bulletin of Volcanology v. 54, p. 484-493.

98120 Cole, J.W., 1969, Gariboldi volcanic complex, Ethiopia: Bulletin of Volcanology v. 33, p. 566–578.

98572 Bohrson, W.A., and Reid, M.R., 1997, Genesis of silicic peralkaline volcanic rocks in an ocean island setting by crustal melting and open-system processes: Socorro Island, Mexico: Journal of Petrology v. 38, p. 1137.

98654 Bacon, C.R., and Druitt, T.H., 1988, Compositional evolution of the zoned calcalkaline magma chamber of Mount Mazama, Crater Lake, Oregon: Contributions to Mineralogy and Petrology v. 98, p. 224–256.

98993 Vankova, V., Vanek, J., Hanus, V., Fiala, J., and Chalupsky, V., 1993, Geochemistry of calc-alkaline volcanic rocks in relation to deep structure of Ecuador and southern Colombia: Global tectonics and metallogeny v. 4, p. 113– 155.

99018 Dias, N.A., Matias, L., Louren{ç}o, N., Madeira, J., Carrilho, F., and Gaspar, J.L., 2007, Crustal seismic velocity structure near Faial and Pico Islands (Azores), from local earthquake tomography: Tectonophysics v. 445, p. 301–317.

100041 Horn, S., and Schmincke, H.U., 2000, Volatile emission during the eruption of Baitoushan Volcano (China/North Korea) ca. 969 AD: Bulletin of Volcanology v. 61, p. 537–555.

275