1

GSA Data Repository 2018138

Levy et al., 2018, Mechanics of fault reactivation before, during, and after the 2015 eruption of

Axial Seamount: Geology, https://doi.org/10.1130/G39978.1.

DATA AND METHODS 1

Earthquake Location and Magnitude Procedures 2

Earthquakes were identified and located using data from seven three-component seismic stations 3

deployed as part of the Ocean Observatories Initiative (OOI) Cabled Array (Kelley et al, 2014). 4

An STA/LTA algorithm operating on a band-passed (750 Hz) vertical channel waveform was 5

used to detect P-wave arrivals that were associated to form an initial set of hypocenters. S-waves 6

were then identified using a kurtosis algorithm (e.g., Baillard et al., 2014) applied to the vector 7

sum of horizontal components. Detections of whale calls and other in-water sources were 8

excluded based on the coherence between the seismic waveforms and co-located hydrophone 9

sensors. The remaining P- and S-wave detections were associated to form a catalog of 41,522 10

hypocenters, which were then located using a 1-D velocity model derived from multi-channel 11

seismic studies (Arnulf et al., 2014) and the generalized earthquake-location (GENLOC) 12

package (Pavlis et al., 2004). The earthquake hypocenters were relocated using the double-13

difference algorithm hypoDD (Waldhauser and Ellsworth, 2000). A final catalog of 19,049 14

earthquakes, each with at least nine defining phases, was derived from processing differential 15

2

arrival times using an 18-step iterative conjugate gradient method with the parameters listed in 16

Table DR1. The median hypocenter location change after relocation is 230 m and the mean of 17

the residuals is 0.029s. Median relative location uncertainty was estimated to be 47 m in the 18

horizontal and 103 m in the vertical direction. 19

Microearthquake moments (Mo) were determined in the frequency domain from P and S 20

arrivals (Brune, 1970): 21

22

where is the crustal density at the hypocenter (2700 kg/m3), v is the average seismic velocity 23

for P (5.09 km/s) or S (2.78 km/s) waves (Arnulf et al., 2014),, r is the slant range from the 24

earthquake hypocenter to the station, K is the free-surface correction taken to be 1.5 and 1.7 for P 25

and S waves, and R is the radiation function assumed to be 0.42 and 0.59 for P or S waves, 26

respectively (Toomey et al., 1985; Weekly et al., 2013). Ωo is the low frequency level of the 27

amplitude spectrum calculated using a Boatwright (1980) model of the form: 28

29

The quality factor (Q), corner frequency (fc) and Ω0 are found simultaneously using a non-linear 30

least squares procedure, and the spectral amplitudes Ω(f) are obtained using a multi-taper (7) 31

approach with a time-bandwidth product of 4. Moment magnitudes (Mw) were estimate from the 32

seismic moments in Nm as (Hanks and Kanamori, 1979): 33

Mw= 2/3 log10(Mo) – 6 34

35

Mo4v3r

o

RK

( f )

oe( ft/Q)

[1( f / fc)4 ]1/2

3

Table DR1. HypoDD Parameters 36

Parameters used in the hypoDD hypocenter relocation process. The max distance between linked 37

pairs and residual threshold became progressively smaller as iterations progressed through the 38

relocation routine in order to minimize and remove outliers. 39

Iterations P weight S weight Max link

separation (km)

Factor multiplied by the

standard deviation of

the residuals

Damping

1‐6 1 0.5 3 8 240

7‐12 1 0.5 2 7 240

13‐18 1 0.5 1 6 240

40

41

Composite Focal Mechanisms 42

Composite focal mechanisms are commonly used to decrease solution uncertainty when the 43

number of recording stations is limited and/or the signal-to-noise ratio of the arrivals is low (e.g., 44

Waldhauser and Tolstoy, 2011). Cross-correlating the vertical channel waveforms recorded on 45

station AXEC1 identified groups of similar earthquakes. A correlation window of 1.4 s was used 46

starting 0.1 sec before the P wave arrival in order to capture both the P- and S-wave energy. 47

Event clusters were defined based on similarity between waveforms (Fig. DR1) using a 48

hierarchical approach with a correlation coefficient cutoff of 0.7. These data were subset to select 49

clusters with earthquakes spanning a limited time range, resulting in a set of 100 clusters that 50

each contain between 3 and 9 earthquakes. The clusters were then used to track the evolution of 51

fault slip through time. The median time spanned by events within an individual composite is 52

29.3 days; however, this varied based on the event rate throughout the volcanic cycle (22.0 days 53

pre-eruption, 1.2 days during the eruption, and 138.0 days post-eruption) (Fig. DR2). The 54

average horizontal radius of the events within a composite cluster is 212 m. 55

4

P-wave polarities were reviewed by the author for the suite of 501 origins belonging to 56

the composite groups (Levy, 2017). S/P amplitude data were measured in the 3-20 Hz frequency 57

band using the vector sum of the peak amplitudes on each of the three channels, following the 58

procedure of Yang et al. (2012). Focal mechanism solutions were found using HASH 59

(Hardebeck & Shearer, 2002). The best solution was found at the average cluster location, but a 60

set of acceptable focal mechanisms was calculated using an allowed polarity misfit of 10%, an 61

epicenter uncertainty of 50 m, a vertical uncertainty of 100 m, and small variations in the 62

velocity model. The preferred focal mechanism was then found by averaging the acceptable focal 63

mechanisms after the removal of outliers. 64

Solutions run with and without S/P information returned similar nodal plane geometries; 65

however, when the amplitude ratio data were included the average nodal plane uncertainty 66

(NPU) decreased from 38 to 21 and the number of A and B quality (Hardebeck & Shearer, 67

2002) solutions increased from 25 to 81. To further assess the sensitivity of the solution to 68

uncertainty in focal depth, focal mechanism parameters were analyzed for a range of depths 69

between 0.7 and 2.0 km, in 0.1 km increments. The median nodal plane rotation is <24 across 70

this depth range, with fault plane uncertainty typically minimized near the mean centroid depth 71

of earthquakes within the cluster. 72

The mean strike and dip of the inferred fault planes (taken to be the outward dipping 73

nodal plane returned for each solution) were estimated using the vector statistics package of 74

Jones (2005); uncertainty in the mean was found using the 95th percentile of the chi-squared 75

distribution. For the entire dataset the vector mean strike and dip is 34513 and 702, 76

respectively. The strike and dip vector means were also calculated during the identifiable phases 77

5

of the volcanic cycle and were found to be 33321/674 (strike/dip for pre-eruption), 78

33815/695 (syn-eruption), and 00727/754 (post-eruption). 79

Analysis of Bottom Pressure Data 80

Bottom-pressure recorder (BPR) data were averaged over 1-minuate intervals and corrected for 81

tides by subtracting the predicted pressure change associated with ocean tides, as estimated using 82

the OSU tidal model (Egbert et al., 1994) implemented within SPOTL (Agnew, 2012). Pressure 83

was converted to water column height assuming a density of 1030 kg/m3. A least-squares linear 84

regression was used to estimate the inflation rates during different phases of the eruption (Fig 85

DR3). 86

Seismic Moments and Cumulative Slip Estimates 87

The seismic moment for each earthquake was found using the median of the individual P and S 88

wave attenuation-corrected spectral amplitudes. The average slip (d) calculation along the 89

eastern ring fault is estimated from the sum of the scalar seismic moments: Mo=μAd. The shear 90

modulus (μ) was taken to be 1-4 GPa, appropriate for fractured basalt in the upper crust 91

(Gudmundsson, 2016), and the area (A) is estimated based on the along-strike (3.0 km) and 92

down-dip (1.75 km) dimensions of the seismically active eastern margin of the caldera. 93

Assuming a 1-4 GPa range of shear modulus, the cumulative slip along the eastern portion of the 94

ring fault was estimated to be 8-30 cm in the three months prior to the eruption, 44-175 cm over 95

the ~1-month period during the eruption, and <0.5 cm over the 19-month period post eruption. 96

These estimates were then related to the differential vertical motion between BPR stations 97

AXCC1 and AXEC2 to constrain the relative coupling between the eastern fault system and the 98

geodetic uplift. 99

6

REFERENCES FOR DATA REPOSITORY 100

Agnew, D. C., 2012, SPOTL: Some Programs for Ocean-Tide Loading, SIO Technical Report, 101

Scripps Institution of Oceanography, http://escholarship.org/uc/item/954322pg. 102

Baillard, C., Crawford, W. C., Ballu, V., Hibert, C., and Mangeney, A., 2014, An automatic 103

kurtosis-based P and S-phase picker designed for local seismic networks. Bulletin 104

Seismological Society of America, v. 104, no.1, p. 394–409 doi:10.1785/0120120347 105

Boatwright J., 1980, A spectral theory for circular seismic sources: simple estimates of source 106

dimension, dynamic stress drop and radiated energy. Bulletin Seismological Society 107

America 70, 1–27 108

Brune, J. N., 1970, Tectonic stress and the spectra of seismic shear waves from earthquakes, 109

Journal of Geophysical Research 75, 4997–5009 110

Egbert, G. D., A. F. Bennett, and M. G. G. Foreman, 1994, TOPEX/POSEIDON tides estimated 111

using a global inverse model, Journal of Geophysical Research., v. 99, no. 24, p. 24,821–112

24,852. 113

Gudmundsson et al., 2016, Gradual caldera collapse at Bárdarbunga volcano, Iceland, regulated 114

by lateral magma outflow, Science 353, doi:10.1126/science.aaf8988 115

Hanks, T. C., and Kanamori, H., 1979, Moment magnitude scale, Journal of Geophysical 116

Research, 84, 2348–2350 117

Hardebeck, J. L., and Shearer, P. M., 2002, A new method for determining first-motion focal 118

mechanisms. Bulletin Seismological Society of America, v. 92, no. 6, p. 2264–2276. 119

doi:10.1785/0120010200 120

Hardebeck, J. L., and Shearer, P. M., 2003, Using S/P amplitude ratios to constrain the focal 121

mechanisms of small earthquakes. Bulletin of the Seismological Society of America, v. 93, 122

no. 6 p. 2434–2444. doi:10.1785/0120020236 123

Jones, T. A., 2005, MATLAB functions to analyze directional (azimuthal) data-I: Single-Sample 124

inference. Computers and Geosciences, v. 36, no. 4, p. 520–525. 125

doi:10.1016/j.cageo.2009.07.011 126

Levy, S., Bohnenstiehl D.R., Wegmann, K., Byrne, P., 2017, Ring Fault Mechanics Surrounding 127

the 2015 Eruption of Axial Seamount, [Master’s thesis] Raleigh, North Carolina State 128

University, North Carolina. 129

Pavlis, G. L., F. L. Vernon, D. Harvey, and D. Quinlan, 2004, The generalized earthquake 130

location (GENLOC) package: A modern earthquake location library, Computers in 131

Geosciences, 30, 1079-1091. 132

Toomey, D. R., Solomon, S. C., Purdy, G. M. and Murray, M. H., 1985, Microearthquakes 133

beneath the median valley of the Mid-Atlantic Ridge near 23N: Hypocenters and focal 134

mechanisms, Journal of Geophysical Research, 90, 5443–5458. 135

7

Waldhauser, F., and Ellsworth, W. L., 2000, A Double-difference Earthquake location algorithm: 136

Method and application to the Northern Hayward Fault, California. Bulletin of the 137

Seismological Society of America, v. 90, no. 6, p. 1353–1368. doi:10.1785/0120000006 138

Waldhauser, F., and Tolstoy, M., 2011, Seismogenic structure and processes associated with 139

magma inflation and hydrothermal circulation beneath the East Pacific Rise at 950’N. 140

Geochemistry, Geophysics, Geosystems, v. 12, no. 9, p. 1-18. doi:10.1029/2011GC003568 141

Weekly, R. T., Wilcock, W. S. D., Hooft, E. E. E., Toomey, D. R., McGill, P. R., and Stakes, D. 142

S., 2013, Termination of a 6 year ridge-spreading event observed using a seafloor seismic 143

network on the Endeavour Segment, Juan de Fuca Ridge. Geochemistry, Geophysics, 144

Geosystems, v. 14, no. 5, p. 1375–1398. doi:10.1002/ggge.20105 145

Wilcock, W. S. D., Tolstoy, M., Waldhauser, F., Gracia, C., Tan, Y. J., Bohnenstiehl, D. R., 146

Caplan-Auerbach, J., Dziak, R.P, Arnulf, A.F., and Mann, E. M., 2016, Seismic Constraints 147

on caldera dynamics from the 2015 Axial Seamount eruption. Science, v. 354, no. 6318, p. 148

1395-1399. doi:10.1126/science.aah5563 149

Yang, W., Hauksson, E., and Shearer, P. M., 2012, Computing a large refined catalog of focal 150

mechanisms for southern California (1981-2010): Temporal stability of the style of faulting. 151

Bulletin of the Seismological Society of America, v. 102, no. 3, p. 1179–1194. 152

doi:10.1785/0120110311 153

154

155

8

FIGURE CAPTIONS: 156

Figure DR1. a) Vertical channel waveforms for similar events in the cluster, used to generate a 157

composite focal mechanism. b) Vertical and the horizontal waveforms for a single arrival shown 158

with windows (boxes) used to estimate the noise, P-arrival, and the S-arrival amplitudes. c) 159

Composite focal mechanism with red and blue indicating upward and downward polarity, 160

respectively, and the size of the circle scaled to represent the S/P amplitude. 161

162

Figure DR2. Graph showing the length of time spanned by events in each composite focal 163

mechanism cluster. The y-axis represents the cluster numbered from 0 to 100. Each horizontal 164

line begins and ends at the times of the earliest and latest earthquake in a cluster. Background 165

shading denotes phases of volcanic inflation and deflation. 166

167

Figure DR3. Time series of elevation data from BPR stations (a) AXCC1, (b) AXEC2 and (c) 168

AXID1 processed to remove tidal variations. Slopes indicating inflation rates are shown by 169

dashed black lines. Background shading denotes phases of volcanic inflation and deflation. 170

AXAS1

AXAS2

AXCC1

AXEC1

AXEC2

AXEC3

AXID1

432421

433314

424688

432421

433314

424688

432421

433314

424688

432421

433314

424688

432421

433314

424688

432421

433314

424688

432421

433314

<1

5

10

Down

Up

Average S/P Ratio Polarity

AXEC1

432421

0 1 2 3

432421

432421

0 1 2 3 4Seconds

0 1 2 3 4Seconds

0 1 2 3 4Seconds

Seconds4

VerticalNorthEast

a)

b)

c)

Fig. DR1

Jan’15 Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan’170

10

20

30

40

50

60

70

80

90

100Co

mpo

site

Foc

al M

echa

nism

Clu

ster

#

Eruption

Slower Ination Rate

Post-Eruption Pre-eruption Post-eruption

Date (MMM-YY)

Fig. DR2

Slower Ination RateEruption

40.7cm/yr

91.6 cm/yr

60.9 cm/yr

20.2 cm/year42.6 cm/year

30.3 cm/year

Dec’14 Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec’160

1

2

3

21.1 cm/year

28.9 cm/year15.4 cm/year

Dec’14 Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec’160

1

2

3

Dec’14 Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec’160

1

2

3

AXCC1

AXEC2

AXID1

Chan

ge in

ele

vatio

n (m

)Ch

ange

in e

leva

tion

(m)

Chan

ge in

ele

vatio

n (m

)

a)

b)

c)

Fig. DR3