+ All Categories
Home > Documents > Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1...

Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1...

Date post: 28-Feb-2021
Category:
Upload: others
View: 5 times
Download: 0 times
Share this document with a friend
50
Energy and Rupture Dynamics are Different for 1 Earthquakes on Mature Faults vs. Immature Faults 2 Rebecca M. Harrington Karlsruhe Institute of Technology Geophysical Institute [email protected] Emily E. Brodsky University of California, Santa Cruz Dept. Earth Sciences [email protected] 3 May 3, 2011 4 1
Transcript
Page 1: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Energy and Rupture Dynamics are Different for1

Earthquakes on Mature Faults vs. Immature Faults2

Rebecca M. HarringtonKarlsruhe Institute of Technology

Geophysical Institute

[email protected]

Emily E. BrodskyUniversity of California, Santa Cruz

Dept. Earth Sciences

[email protected]

3

May 3, 20114

1

Page 2: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Abstract5

We compare radiated energy per unit moment, and the seismologically ob-6

servable fracture energy per unit moment between earthquakes on immature7

fault surfaces in Parkfield, California, and Mount St. Helens Volcano, and on8

a mature fault surface, namely the San Andreas fault near Parkfield, Califor-9

nia. A comparison between the two populations indicates that earthquakes on10

immature fault surfaces exhibit more self-similar behavior. Energy to moment11

ratios remain roughly constant for immature faults, and decrease with magni-12

tude for mature fault surfaces. The decrease in the parameters for the mature13

population results from an unchanging fault area for earthquakes with catalog14

magnitudes smaller that M 3.0 on the San Andreas fault. A comparison of15

energy-moment scaling with other studies suggests that the proposed differ-16

ences in energy scaling between mature and immature faults is consistent with17

previously published data sets from mature faults in California, and immature18

faults in deep mines. In addition, the observed energy values fall within the19

expected range. The difference in source parameter scaling between immature20

and mature populations suggests that ordinary scaling relationships relating21

moment and area and an assumed constant stress drop are not valid for every22

earthquake population individually.23

2

Page 3: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

1 Introduction24

Seismically observable earthquake source parameters such as duration (τ), seismic25

moment (M0) and the radiated seismic energy (ER) provide the only direct clues26

as to how faults rupture [Wesnousky , 2006; Kanamori and Rivera, 2004; Choy and27

Kirby , 2004; Wyss and Brune, 1968]. Quantities such as the seismic moment, and the28

radiated seismic energy can be measured directly from the earthquake source spec-29

trum. Other quantities, such as static stress drop (∆σ), or the seismically observable30

fracture energy (EG) must be derived via theoretical relationships related to rupture31

models. The scaling of earthquake source parameters with size provides clues as to32

whether rupture processes are independent of earthquake size, i.e. are scale invari-33

ant. A number of observations suggest that some parameters, such as static stress34

drop and the ratio of radiated seismic energy to moment (ER

M0), do not vary with size,35

leading to assumptions of scale invariance over a large magnitude range [Kanamori36

and Brodsky , 2004; Kanamori and Anderson, 1975; Abercrombie, 1995; Ide et al.,37

2003; Imanishi and Ellsworth, 2006; Abercrombie and Rice, 2005; Prieto et al., 2004;38

Shearer et al., 2006].39

The accepted values of roughly constant static stress drop range over 2 to 3 orders40

of magnitude for small earthquakes, calling into question the assumption of constant41

source scaling. Observations of a decreasing ER

M0ratio for small-magnitude earthquakes42

also seem to contradict the constant scaling assumption [Abercrombie, 1995; Prejean43

and Ellsworth, 2001; Venkataraman and Kanamori , 2004a; Mayeda et al., 2007]. The44

difficulties in accurately measuring radiated seismic energy for small earthquakes45

unfortunately makes the scaling issue even more ambiguous [Venkataraman et al.,46

2006b]. Certainly some of the scatter in measured energy values results from difficul-47

3

Page 4: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

ties in accurately correcting for attenuation, as well as difficulties in accounting for48

all of the energy due to bandwidth limitations [Venkataraman et al., 2006a; Ide and49

Beroza, 2001]. However, some work suggests that heterogeneous properties in the50

faulting environment might influence the scaling of parameters such as ∆σ, and ER

M0.51

Such observations suggest that geophysical factors might also account for some of the52

range in scaling observations [Venkataraman and Kanamori , 2004a; Choy and Kirby ,53

2004]. One such physical factor which may contribute to the range in source obser-54

vations is fault surface roughness, which is modulated by fault maturity (cumulative55

displacement).56

Recent studies using laser techniques to image fault surfaces indicate that they57

become smoother with more cumulative displacement. Observations at the cm - km58

scale suggest that, similar to field observations at the km - 100 km scale, immature59

fault surfaces with less cumulative displacement are rougher than more mature faults60

(with more cumulative displacement) [Sagy et al., 2007; Sagy and Brodsky , 2009;61

Wesnousky , 1988]. The results of Sagy et al. [2007]; Sagy and Brodsky [2009] suggest62

that immature fault surfaces may be assumed self-affine in the direction of slip, i.e.63

surface roughness is scale invariant. However, mature fault surfaces become more64

smooth at smaller scales, and develop quasi-elliptical geometric asperities on the65

order of 10’s of meters for faults with cumulative slip on the order of 100’s of meters66

[Sagy and Brodsky , 2009]. Given the physical differences in surface geometry between67

mature and immature faults, it follows that earthquakes might rupture differently,68

depending on the faulting environment in which they occur.69

Ohnaka [2003] suggests that there are observable differences in dynamic source70

properties between rough and smooth faults in laboratory experiments. He develops71

4

Page 5: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

a constitutive relation dependent on fault friction and seismologically observable frac-72

ture energy that governs fault rupture. The relation relates seismologically observable73

fracture energy to fault surface roughness, suggesting that the energy partitioning in74

an earthquake (and therefore the amount of radiated energy) depends on surface75

roughness as well. In this study, we examine whether dynamic properties between76

rough and smooth faults differ on a larger scale, by using a new compilation data set77

of earthquake recordings. The assumption that roughness correlates with maturity78

implies that the comparison between rough and smooth faults is also a comparison79

of how maturity affects the dynamic properties of earthquakes.80

Harrington and Brodsky [2009] showed that a group of earthquakes on the San81

Andreas fault have a duration-seismic moment scaling suggestive of an approximately82

constant fault area. Here we present new evidence using an expanded data set to sug-83

gest that fault maturity affects the seismically radiated energy (ER), the energy radi-84

ated away from the fault in an earthquake [Kanamori and Heaton, 2000; Abercrombie85

and Rice, 2005; Kanamori and Brodsky , 2004]. By comparing the energy-moment86

ratios (ER

M0, EG

M0) of earthquakes on the San Andreas fault to those on secondary faults87

in the Parkfield area, and on new fault surfaces in the Mount St. Helens edifice,88

we will show that the earthquakes on immature faults obey constant energy scaling,89

while those on mature faults exhibit a size-dependent energy scaling. In addition,90

we will present our results in the context of other studies, to show the more general91

features of our observations. Additionally, we will offer an interpretation of energy92

partitioning in the context of our observed energy scaling, as well as other studies of93

fault roughness and damage.94

The paper begins with the directly measured parameters: seismic moment and cor-95

5

Page 6: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

ner frequency. After reviewing the methodology and geological context, we present96

measurements showing distinct trends in corner frequency and moment for mature and97

immature faults. We then proceed to interpret these results in terms of energy. We re-98

view a slip-weakening rupture model to infer energy terms from the observations, and99

then show how the ratios of radiated seismic energy to moment, and seismologically100

observable fracture energy to moment (ER

M0, EG

M0) for the mature population deviate101

from the constant scaling expected from the model based considerations. Finally,102

we compare our observations with studies which measure similar source properties in103

other locations, in order to show that they are a general feature.104

2 Earthquake Populations105

Making a comparison of dynamic source parameters between mature and immature106

earthquake populations requires a definition of maturity. A number of observations107

suggest that differences in dynamic source properties exist between earthquakes oc-108

curring on plate boundaries (interplate), and within plates (intraplate). Such com-109

parisons are based on proximity to the plate boundary, and do not directly relate110

source scaling to cumulative slip [Choy and Kirby , 2004; Choy et al., 2006; Kanamori111

and Anderson, 1975; Zhuo and Kanamori , 1987]. For example, subduction zone112

earthquakes are typically observed to have lower stress drop and radiated energy113

values [Allmann and Shearer , 2009; Venkataraman and Kanamori , 2004a; Choy and114

Boatwright , 2009]. The favored interpretation suggests that subduction zone events115

differ from other types of earthquakes because they occur in a mature faulting envi-116

ronment where the largest amounts of cumulative slip have accumulated. For faults117

6

Page 7: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

which are not exposed at the Earth’s surface, the cumulative displacement on the118

fault surface serves as a measurable proxy for fault roughness. We therefore use it119

here to distinguish mature and immature faults, in order to make a comparison of120

similar sized earthquakes in both environments.121

Observational evidence from roughness measurements of exposed fault surfaces122

suggest that the transitional value of cumulative displacement separating immature123

from mature faults is ∼100-1000 m [Sagy and Brodsky , 2009; Sagy et al., 2007; Savage124

and Brodsky , 2011]. Observations of both mapped fault trace features, and appar-125

ent stress measurements (or the radiated energy-moment ratio) suggest that mature126

faults constitute those accommodating a few kilometers or more of slip [Choy and127

Kirby , 2004; Choy et al., 2006; Malagnini et al., 2010; Wesnousky , 1988]. Therefore,128

we classify the events in our data set occurring on faults with displacement values129

less than 1 km as immature. The study analyzes seismograms from two areas: the130

Parkfield region, and the edifice of Mount St. Helens volcano. Both areas have dense131

enough instrumentation that direct analysis of the waveforms of small earthquakes132

is possible, and all earthquakes in this study have magnitudes less than 3.7. The133

paper separates these earthquakes into two populations: (1) earthquakes on a mature134

fault (namely, the San Andreas fault near Parkfield), and (2) earthquakes on imma-135

ture faults which meet the < 1 km displacement criteria (namely, secondary faults in136

Parkfield and faults in the edifice of Mount St. Helens).137

7

Page 8: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

2.1 Mature Fault Earthquakes: San Andreas Fault at Park-138

field139

Nearly all of the earthquakes in Parkfield listed here are contained in the relocated140

catalog of Thurber et al. [2006] and/or the ANSS catalog (for events in 2006 and 2007141

post-dating publication of the relocated catalog) [Thurber et al., 2006; Waldhauser142

and Ellsworth, 2000, 2002] (Table 1). We use waveforms from the High Resolution143

Seismic Network (HRSN) stations (Figure 1). The HRSN consists of 13 3-component144

borehole stations with depths ranging from 63-572 m, an average depth of 236 m,145

and a sampling rate of 250 sps. The high-quality data permits unusually clear source146

spectral estimation for small events.147

The San Andreas fault near Parkfield is definitively mature with cumulative dis-148

placement values in the range of 300 km [Revenaugh and Reasoner , 1997]. Earth-149

quakes that occur on the active fault strand may therefore be classified in the mature150

population. Of the cataloged earthquakes, 25 events located on the San Andreas fault151

with magnitude values ranging from 1.2 to 3.9 and occur in sufficiently close clusters,152

and at similar source-station distances to the immature events to permit an empirical153

Green’s function analysis. We analyze additional events clustered on the active fault154

in the middle of the HRSN array to verify that we obtain the same quality of results155

for the events not recorded with complete azimuthal coverage.156

Given that displacement often migrates to newly activated strands in a mature157

fault zone, the assumption of a 300 km offset on the active strand may be an overes-158

timate [Thurber et al., 2006; Simpson et al., 2006; Savage and Brodsky , 2011]. Events159

from the mature population may occur on the youngest, active strand, which has160

likely sustained less offset than the cumulative total. Studies of SAFOD fault cores161

8

Page 9: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

suggest that the currently active strand has accommodated well over 100 km of offset162

[Titus et al., 2005; Schleicher et al., 2010]. Geological and geodetic data therefore163

suggest that the youngest fault strands in the area may easily exceed the ∼1 km164

displacement criterion designating a fault mature.165

2.2 Immature Fault Earthquakes: Mount St. Helens Edifice166

and Secondary Faults at Parkfield167

The earthquakes at Mount St. Helens (MSH) consist of 40 events on a newly formed168

fault surface created during the 2004 dome-building eruption (Figure 2). The events169

are located at the base of an extruding solid rock spine in the crater of the volcano in170

February and March of 2005, when the primary station used in our analysis, MIDE,171

was operational (seismic moment and corner frequency values listed in Table 2). Mo-172

ment magnitude calculations ranging from 0.4 to 1.4 result from using spectral ratios173

multiplied by the long period amplitude of the earthquake spectra for the empirical174

Green’s function event used at MSH (detailed in the seismic moments and magnitudes175

section). The source time functions used here all result from spectral deconvolution176

of a single empirical Green’s function. We therefore infer that relative moment values177

are precise, and that errors in absolute moment stem primarily from the moment178

estimation of the empirical Green’s function. The earthquakes at Mount St. Helens179

are not cataloged, and therefore not located, as the numerous events resulting from180

the ongoing eruption in 2005 prevented many events from being analyzed. We find181

co-located earthquakes pairs by cross-correlation of events over both low- and high-182

gain channels of Pacific Northwest Seismographic Network (PNSN) stations MIDE183

and NED. Only two seismic stations were both simultaneously operational in Febru-184

9

Page 10: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

ary and March of 2005, and located close enough to the crater to record the small185

earthquakes associated with the solid rock spine extrusion in a large enough dynamic186

range necessary for our analysis. Stations MIDE and NED are short-period, vertical187

component seismometers peaked at 1 Hz, and sample rates of 100 sps. Additional188

stations on the volcanic edifice recorded larger events, and were used to estimate the189

shear wave velocity at shallow depths (Figure 2).190

Earthquakes more than 5 km from the main trace of the San Andreas fault at191

Parkfield are inferred to be on secondary faults, and of these, 11 are sufficiently192

clustered to permit analysis (Figure 1). These events occur well outside of the San193

Andreas system on small faults with virtually no expression of a surface trace. The194

three easternmost earthquakes in Parkfield do occur in the vicinity of the Kettleman195

Hills fault (∼ 5km) [Lin and Stein, 2006]. However, we do not exclude small events196

that happen on larger faults from the immature population, as long as the total fault197

displacement is less than 1 km. In that context, we argue that the three easternmost198

events may also be classified as immature, as the displacement on the faults in the area199

is orders of magnitude less than that on any active strand of the SAF, and likely within200

the 100s of meters range according to cross-sectional plots of the sedimentary structure201

[Lin and Stein, 2006]. The remaining immature events shown in Figure 1 occur202

on secondary faults at distances 30 km or more from the Coalinga, New Idria and203

Kettleman Hills faults. We therefore infer from the large distance that the remaining204

events are not associated with any of the previous large earthquakes that occurred on205

the larger faults (distance cited is perpendicular from strike) [Lin and Stein, 2006].206

The immature events at Parkfield and Mount St. Helens have different types of207

focal mechanisms and occur in different lithological settings. While most of the events208

10

Page 11: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

on the San Andreas fault have strike-slip focal mechanisms, seven of the 11 events in209

the immature population in Parkfield have reverse faulting focal mechanisms. Focal210

mechanism solutions do not exist for the earthquakes recorded at Mount St. Helens,211

due to the small magnitudes. Some previous studies have found differences in source212

properties correlated with varying faulting style, implying that faulting style may213

influence source scaling characteristics [Allmann and Shearer , 2009; Venkataraman214

and Kanamori , 2004a]. Here we take a different approach by examining the behavior215

as a function of a less-commonly studied parameter: maturity. If systematics are216

revealed as a function of maturity, then the accompanying physical properties of the217

fault, e.g., roughness, are at least as viable an explanation for differences in behavior218

as any previously studied.219

3 Source spectral calculation and fitting220

Our analysis of dynamic source parameters is based on two fundamental parts: es-221

timation of seismic moment and spectral corner frequency using direct observations222

from the spectra, and energy calculations based on a slip weakening rupture model.223

We detail the direct source spectral observations in this section, and continue with224

the more model dependent energy parameters in section 4.225

Accounting for high-frequency attenuation (particularly in the volcanic edifice) is226

crucial to investigating source parameter scaling, particularly for small earthquakes227

where accurate corner frequency estimations are crucial. We use an empirical Green’s228

function approach to eliminate attenuation effects caused by the path the seismic229

waves travel from the source to the receiver, as well as the site effects at the station,230

11

Page 12: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

and instrument response. In both Parkfield and at Mount St. Helens, we use co-231

located event pairs with a magnitude unit or more difference in size to deconvolve232

the path, site, and instrument response from earthquake recordings. The event co-233

location requirement insures a nearly identical source-receiver geometry, permitting234

instrument, and travel path attenuation effects to cancel effectively. The magnitude235

difference requirement permits the assumption that the source-time function of the236

smaller event is sufficiently close to a delta function over the bandwidth of interest.237

The resulting time series after the deconvolution step consists of the earthquake source238

time function. The source time function indicates moment rate of the earthquake as239

a function of time isolated from the other distorting effects listed above. Using the240

source time functions, we calculate the source time function spectra, and estimate241

the the long-period spectral ratio, and spectral corner frequency.242

First, we deconvolve an empirical Green’s function (eGf) from the entire waveform243

of the larger co-located event in order to obtain the source-time function. Figure 3244

shows two examples of eGf’s and their larger respective co-located events from both245

Parkfield and Mount St. Helens (left and right columns respectively). The example246

shows the Parkfield event recorded on the three-components of station VCAB, and247

the Mount St. Helens event on the single component station MIDE. Calculations of248

the source time functions for the Mount St. Helens events result from a water-level249

deconvolution (spectral division) of the eGf from the larger event. We calculate the250

source-time functions of the Parkfield events using a projected Landweber deconvo-251

lution method. The method is a regularizing, iterative approach in the time domain252

that enforces a positivity constraint on the solution. The approach is outlined in253

detail in Lanza et al. [1999]. As a check on the method, we compare the Landweber254

12

Page 13: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

source time functions to those obtained with a spectral deconvolution, finding nearly255

identical results (Figure 4). The Landweber approach has the added advantage of256

expanding the frequency band over which the signal to noise spectral ratio exceeded257

a factor of two for the Parkfield data, where the source-receiver distances are much258

larger than those at Mount St. Helens. Figure 4 shows the source time functions for259

each of the event pairs depicted in Figure 3.260

Second, we calculate the source spectra of each event using a multi-taper spectral261

calculation. The spectra are shown in the right column of Figure 4. The dashed262

lines represent the least-squares fit to a Brune spectral model, and the solid gray263

lines represent the noise spectra (Figure 4). Noise spectra were calculated from a264

time window directly preceding the source pulse. We used time windows of 256 data265

points, and 128 data points for the Parkfield and Mount St. Helens data respectively,266

corresponding to a time window of ≈ 1 second, for both the source-time function and267

noise spectral calculations.268

We calculate the Parkfield earthquake source spectra using the RMS spectra de-269

termined using the individual station component spectra. We stack the available270

source time spectra determined at each station in order to have a single, spatially271

averaged source time function spectrum for each event. In addition to the co-location272

and magnitude criteria, we choose events in the mature populations such that events273

located north and south of the HRSN array have source-receiver distances similar to274

events occurring on secondary faults. We also choose events in the middle of the array275

to check that we obtain similar source time function calculations when azimuthal cov-276

erage differs. For full details of the source time function spectral calculation, we refer277

the reader to Harrington and Brodsky [2007] and Harrington and Brodsky [2009].278

13

Page 14: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

We model the source spectra for each event using a Brune spectral model with279

spectral falloff n = 2 [Brune, 1970; Abercrombie, 1995].280

Ω(f) =Ω0

(1 + ( ffc

)n)(1)281

We calculate a least-squares fit to the spectral data using the Brune spectral model282

in order to calculate the values for the corner frequency (fc) and the long period283

spectral amplitude (Ω0). The attenuation term typically included in Equation 1 is284

omitted here, as the spectra have already been attenuation corrected through the eGf285

deconvolution.286

3.1 Seismic moments and magnitudes287

Slightly different methods are appropriate for determining the seismic moment from288

the earthquake spectra in each location, due to the difference in earthquake relative289

locations and the available catalogs. We therefore describe the moment calculation290

separately for Parkfield and Mount St. Helens in the subsequent two subsections.291

3.1.1 Mount St. Helens292

The events at Mount St. Helens fall below the catalog completeness threshold, and293

are not cataloged. We calculate the moment of our eGf event using the long period294

amplitude of the earthquake spectrum, calculated using a time window of 128 samples295

starting 28 samples (i.e. 28 msec) before the P-wave arrival. We then calculate the296

moment using the long-period spectral amplitude, and the equation [Abercrombie,297

14

Page 15: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

1995; Aki and Richards , 2002]298

M0 = 〈U2φθ〉Rpψ4πρα3rΩ0. (2)299

where ρ is the rock density (2.5×103 kg/m3), α is the P-wave velocity (1400 ms

), r300

is the distance to the station, 〈Uφθ〉 is the radiation pattern coefficient, and Rpψ is301

the free surface correction [Aki and Richards , 2002]. Although the events are not302

located, we can infer r based on the distance from the station to the active extruding303

spine. However, the events are shallow, and the source depths relative to the station304

are poorly constrained on the steep mountain. We therefore omit the free surface305

correction in the moment calculation for the events at Mount St. Helens, as we have306

no way of confidently estimating the incidence angle. Section 4.1.1 details the origin307

of the P-wave velocity value. The eGf event used for 36 out of our 40 events was308

extremely small (estimated magnitude of 0.07 ± 0.01 ), and therefore only recorded309

on a single station located an estimated ∼150 m from the source. We calculate310

the moments of the family of 36 events using the long period spectral amplitude311

of the eGf multiplied by the spectral ratios. The moment and corner frequency for312

the remaining four events recorded at station NED are calculated similarly to those313

events recorded at MIDE. We obtain a quantitative error estimation by applying314

the approach described in Prieto et al. [2007] for a single station to the individual315

deconvolved spectra. We estimate the errors of the source parameters using the316

standard deviation of a sub-set of parameter estimations. The sub-set is created from317

the single record by removing one of the Slepian tapers, and repeating the spectral318

estimation omitting a different taper for each estimation [Park et al., 1987]. The319

15

Page 16: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

procedure varies slightly when multiple stations are available, in that the sub-set of320

parameter estimations is created using the multiple station recordings, rather than the321

multiple tapers. We refer the reader to Prieto et al. [2007] for a detailed description.322

3.1.2 Parkfield323

The earthquakes in Parkfield are not clustered at a single location, as are those at324

Mount St. Helens. Each source time function requires a separate empirical Green’s325

function (eGf), making calculating the relative seismic moment from a single event326

impossible. We therefore calculate seismic moments directly from the earthquake dis-327

placement spectra using a two-second window around the S-wave arrival, estimating328

the long period spectral amplitude (Ω0) with a least squares fit to a Brune spectral329

model. We obtain the seismic moment calculation from Ω0 by using Equation 2,330

substituting the shear wave velocity (β) for α, and dividing by the mean radiation331

pattern (Uθφ = 0.63 for S-waves) [Lay and Wallace, 1995]. The hypocentral locations332

of the Parkfield events are known, permitting an estimation of the incidence angle333

of the seismic wave at the station. We therefore include the free surface correction334

(Rpψ) for the Parkfield events.335

We check our work by using the standard relations between M0 and Mw to com-336

pare our calculated Mw and the catalog magnitude values (mainly Mc, with few ML337

values) [Hanks and Kanamori , 1979]. Figure 5 indicates that the Mw values calcu-338

lated here correspond well to catalog values, with some scatter found in the immature339

population. The ratio of Mw : Mc is 0.92 ± 0.04 for the mature (blue) events, and340

1.05 ± 0.13 for the immature events (red, Figure 5). Using the P-wave arrival for the341

Parkfield events instead of the S-wave arrival produces the same trend, however, with342

16

Page 17: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

significantly more scatter in the values. The scatter in the immature event values343

(red diamonds) in Figure 5 likely results from the lack of azimuthal coverage by the344

HRSN stations (Figure 1). We repeated the moment calculations using additional345

waveforms from nearby stations with high-quality recordings in efforts to reduce the346

scatter via improved azimuthal coverage. The red squares indicate the Mw calcula-347

tions using the additional stations (∼5-10 extra stations for each event), suggesting348

that they help reduce the scatter. The moment values calculated with the additional349

stations are therefore used in the analysis.350

We estimate the errors for moment, corner frequency calculations, and energy351

ratios using the multiple station jackknife error estimation described in Prieto et al.352

[2007]. The method is similar to the single station jackknife error estimation described353

above. It works as follows: given a set of N stations, one calculates a parameter (such354

as the spectral corner frequency) from the station averaged spectra. The calculation355

is performed N times, each time with one station removed from the average. The356

standard deviation of the N parameter calculations provides the jackknife error for the357

given parameter. The errors for the energy values are estimated in a similar manner.358

The energy-moment ratios are model based parameters dependent on the moment359

and corner frequency values (discussed in section 4.1.1). We estimate the errors for360

energy values using the N estimations of M0 and fc to calculate N estimations of361

energy ratios, the standard deviation of which constitute the energy errors.362

The stability of the result for a particular choice of eGf is mainly dependent on363

the slight differences in hypocentral location, and the source-station geometry. For364

example, if two events are not perfectly co-located, the effect on the source time365

function will be less apparent on a station with travel path in line with the two366

17

Page 18: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

events compared to a station with travel path normal to the event separation distance.367

Therefore, the multiple station jackknife error estimation is particularly well suited to368

quantify the stability of the source time function solutions, and their resulting spectra369

for our data set in Parkfield. While we do not have multiple stations available for370

the events at Mount St. Helens, given the unique geometry of the extruding spine,371

the short source receiver distance, and the similarity of the waveforms within event372

pairs, we estimate that difference in co-locations would not be greater than a scale of373

meters, and that the solutions are also stable as suggested by the single-station error374

estimation.375

3.2 Results376

Values of seismic moment and corner frequency estimated from the source spectra sug-377

gest that the earthquakes in the immature population exhibit constant stress drop378

scaling, shown by a M0 ∝ f−3c dependence (Figure 6) [Kanamori and Rivera, 2004].379

The earthquakes in the mature population have corner frequencies which remain380

roughly constant, a scaling which is consistent with an roughly constant source dura-381

tion [Harrington and Brodsky , 2009]. Although stress drop values are similar for all382

events analyzed here, the immature population at Mount St. Helens does not follow383

the same line of constant stress drop in Figure 6. The reason for the difference is that384

the shear wave velocity in the volcanic edifice is approximately four times slower than385

in Parkfield, resulting in slower rupture velocities, and subsequently longer rupture386

durations [Harrington and Brodsky , 2007].387

18

Page 19: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

4 Interpretation in Terms of Energy388

Using the spectra along with the modeled M0, and fc values, we can calculate other389

source parameters such as the energy moment ratio (ER

M0), and the seismologically ob-390

servable fracture energy moment ratio (EG

M0). A comparison of all source parameters391

between populations permits us to make inferences regarding how the fault maturity392

affects their relative scaling. We first review the theoretical background for determin-393

ing the energy parameters in the theoretical basis subsection before presenting the394

inferred energy results based on the source spectral observations above.395

4.1 Theoretical basis396

4.1.1 Radiated Energy397

The radiated seismic energy (ER) can be calculated for a given earthquake by inte-398

grating the source velocity spectrum (I) observed at a single station [Venkataraman399

et al., 2002; Venkataraman and Kanamori , 2004b; Prieto et al., 2004].400

I =

∫ f2

f1

2πfΩ0

1 + ( ffc

)n

2

df (3)401

Ω0 is the long-period spectral amplitude observed at a given station, fc is the402

spectral corner frequency, and n is the spectral falloff. f1 and f2 represent the limits403

of integration determined by the bandwidth over which the signal to noise ratio is404

high. Equation 2 relates Ω0 to the M0 via the source-station distance and material405

properties. The radiated seismic energy (ER) in terms of I is then given by:406

ER = 〈U2θφ〉R2

pψ4πρβr2I, (4)407

19

Page 20: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

where 〈U2θφ〉 = 2/5 is the mean S-wave radiation pattern over the focal sphere, Rpψ408

is the free surface correction term, β is the S-wave velocity, ρ is the rock density,409

r is the source-receiver distance, and I is a definite integral given by Equation A.6410

[Abercrombie, 1995; Boatwright and Fletcher , 1984; Aki and Richards , 2002].411

Ide and Beroza [2001] show that radiated seismic energy values are often under-412

estimated due to spectral integration over a limited bandwidth. They suggest that413

radiated energy may be better estimated by integrating the power spectral density414

of a Brune source spectral model from zero to infinite frequency, given that the in-415

tegral has a closed form. Following Ide and Beroza [2001] and using the analytical416

expression resulting from integration of the square of Equation 3 from zero to infinite417

frequency, we determine the analytical expression for the radiated energy in terms of418

the moment and corner frequency obtained from the Brune spectral model (derivation419

in Appendix A):420

ER =π2

5ρβ5M2

0 f3c . (5)421

We calculate the energy values presented here using Equation 5. Equation 5 is422

independent of geometrical spreading and attenuation, and is therefore appropriate423

for source time spectra. The source parameter calculations performed here are based424

on the S-wave values, because the S-wave energy accounts for more than 97% of the425

total radiated energy [Kanamori et al., 1993].426

We use a shear wave velocity value of 3750 m/s given in the velocity model of427

Thurber et al. [2006] for the depth range of our events (namely, 4 to 8.8 km, with428

most events having depths ranging from 5 to 8.5 km, Table 1) in Parkfield. Allmann429

and Shearer [2007] found an apparent static stress drop depth dependence when430

neglecting shear wave velocity depth dependence in Parkfield, suggesting that the431

20

Page 21: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

assumption of a constant shear velocity may be problematic. However, the strongest432

parameter changes they found were at depths above 3 km, and below 12 km. The433

values of shear wave velocity remains fairly constant within the depth range used434

here.435

For Mount St. Helens, we use a value of β = 800 ms

for the loosely consolidated436

volcanic rock and ash in the edifice. The value of 800 ms

results from the upper-437

limit value calculations of the average S-wave velocity at depths similar to the data438

set determined from P-S wave arrival times of earthquakes occurring in the same439

location (and during the same time period). Many of the earthquakes shown here440

were too small to be recorded on many stations, however, a number of larger events441

with clear P- and S-wave arrivals were recorded on additional stations in the volcanic442

edifice. We were therefore able to calculate velocity estimations using the slightly443

larger events with similar travel paths to our data set. (Cataloged events during the444

same time period are also located at the base of the actively extruding spine). Using445

29 larger events and seven stations azimuthally distributed over the volcanic edifice,446

we assume a Poisson solid, and obtain an estimated shear velocity value of 670 ±447

130 m/s. We take the upper-limit as the S-wave velocity value, as it is closer to the448

larger values cited by tomographic studies [Moran et al., 2008]. We use the values449

calculated here rather than values cited by Moran et al. [2008], as the short source-450

receiver distances and the similar locations may represent a more reliable average451

shear velocity value than one obtained from velocity models lacking resolution on the452

scale of 10s of meters.453

21

Page 22: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

4.1.2 Seismologically Observable Fracture Energy454

In addition to the radiated energy and static stress drop, we examine the scaling of455

fracture energy with earthquake size. The relationship between EG and ER can be456

understood by considering a slip-weakening earthquake rupture model which assumes457

no overshoot or undershoot, such as that described by fracture mechanics theory458

[Anderson, 2004]. In such a model, rupture initiates as the stress on the fault reaches459

some yield value. The initial stress on the fault surface drops until the fault slips460

some critical value (Dc), due to plastic yield in a zone near the crack tip [Freund ,461

1979; Kanamori and Heaton, 2000; Kanamori and Brodsky , 2004; Tinti et al., 2005].462

Assuming no under- or overshoot requires that once the slip dependent stress drops463

to the frictional value, σ1 (at D = Dc), sliding continues at the frictional stress level464

until rupture terminates (at D > Dc). Using such a slip-weakening model permits465

one to relate radiated energy and fracture energy in a straightforward way. One can466

show that ER is directly proportional to the sum of stress drop and seismologically467

observable fracture energy EG [Kanamori and Heaton, 2000; Abercrombie and Rice,468

2005; Kanamori and Brodsky , 2004]. Consider that the total energy release in an469

earthquake (∆W ) is partitioned into radiated energy, ER, seismologically observable470

fracture energy, EG, and the constant component of friction, Ef .471

∆W = ER + EG + Ef (6)472

22

Page 23: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Alternatively, ∆W can also be written in terms of the average stress, the fault area,473

and the average slip on the fault,474

∆W =σ1 + σ0

2AD, (7)475

where σ1 and σ0 are the final and initial stresses respectively. Assuming the frictional476

stress (σf ) on the fault surface is equal to the final stress (i.e. no overshoot or477

undershoot) implies the frictional energy can be written in terms of the final stress.478

Ef = σ1AD (8)479

Substituting Equation 8 into the in the Equation 6, produces the following relation:480

1

2∆σAD = ER + EG ⇒ EG =

1

2∆σAD − ER. (9)481

Using the equation for scalar moment,482

M0 = µAD, (10)483

permits writing Equation 9 in terms of the energy moment ratios,484

EGM0

=1

2

∆σ

µ− ERM0

, (11)485

where µ is the rigidity [Keilis-Borok , 1959; Kanamori and Anderson, 1975]. We use486

values of µ = 5× 109 Nm2 , and µ = 3× 1010 N

m2 for Mount St. Helens and Parkfield re-487

spectively, based on the shear velocity in a Poisson solid (µ = ρβ2) [Aki and Richards ,488

23

Page 24: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

2002]. Substituting the analytical value for radiated energy from Equation 5, and the489

expression for the static stress drop on a penny-shaped crack, namely,490

∆σ =7

16M0

(fc

0.32β

)3

, (12)491

we have an expression for EG

M0in terms of seismic moment and corner frequency:492

EGM0

=

(1

2

1

µ

7

16

[1

0.32β

]3

− π2

5ρβ5

)M0f

3c . (13)493

4.2 Energy Results494

Equation 13 indicates that, ER

M0, and EG

M0are both proportional to M0f

3c . Therefore,495

if a group of events has self-similar scaling, a constant ratio for both parameters is496

expected, based on the model assumptions. Deviations from the constant scaling of497

either of these parameters for a particular group of earthquakes would imply that498

scale invariance is not a characteristic of that group of events.499

The trends in energy with earthquake size exhibited in Figure 7 suggest that the500

earthquakes in the mature population do not exhibit scale invariance. The group501

of earthquakes on immature faults, i.e. those on secondary faults in Parkfield, and502

in the Mount St. Helens crater, occur presumably on rougher fault surfaces, based503

on the maturity criterion defined in the introduction, and on fieldwork examining504

the roughness of exposed fault surfaces Sagy et al. [2007]; Sagy and Brodsky [2009].505

Similarly, the events on the San Andreas fault, i.e. the mature population, occur506

presumably on a smoother fault. The ratio ER

M0may be considered as the amount of507

energy radiated per unit fault area per unit slip in an earthquake. We focus on this508

24

Page 25: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

parameter, as it provides a tenuous observational link to the rupture process [Wes-509

nousky , 2006; Kanamori and Rivera, 2004; Choy and Kirby , 2004; Wyss and Brune,510

1968]. If we can assume that differences in fault surface geometry cause the observed511

differences in energy-moment ratio scaling (ER

M0) between immature and mature pop-512

ulations, then the scaling differences demonstrate how faulting environment might513

influence properties of rupture (Figure 7).514

Using a linear regression to fit both the mature and immature populations indi-515

cates a slope of 0.9 ± 0.16 and -0.2 ± 0.08 respectively in the variation of energy-516

moment ratios with earthquake size (Figure 7). The fit suggests a negligible variation517

of the energy ratio for the immature (rougher) fault surfaces, and a much stronger518

dependence of energy on moment for the mature (smoother) fault surface. If we519

consider the case where our shear wave velocity for Mount St. Helens has the mean520

estimated value rather than the upper-limit value (namely, a value of 670 ms

instead521

of 800 ms

), the fitted slope value changes from -0.2 to -0.3 for the immature popula-522

tion, i.e., the radiated energy does not perceptibly change with moment. Even with523

the lower estimate of shear velocity at Mount St. Helens, the energy dependence on524

moment is still significantly greater for the mature population.525

4.2.1 Comparison with faults from the literature526

The earthquakes in both our mature and immature populations occur in non-typical527

faulting environments. The events comprising the mature population occur near528

the creeping/locked transition of the San Andreas fault, and the majority of those529

comprising our immature population occur in a volcanic edifice, rather than a tectonic530

fault. We consider the possibility that our observations result from using earthquakes531

25

Page 26: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

in a unique environment by comparing our energy-moment ratios and static stress532

drop values to those of other studies. The analogous immature events consist of those533

analyzed by a number of mining seismicity studies [Yamada et al., 2007; Kwiatek et al.,534

2011; Gibowicz et al., 1991; Urbancic and Young , 1993; McGarr et al., 2010], while535

the mature events consist of events occurring on faults within the Basin and Range536

faulting system in the western US, as well as those associated with the San Andreas537

[Abercrombie, 1995; Abercrombie and Rice, 2005; Mayeda and Walter , 1996].538

Figure 8 plots our observations shown in Figure 7 along with the results of a539

variety of other studies including both immature and mature populations for com-540

parison [Yamada et al., 2007; Kwiatek et al., 2011; Gibowicz et al., 1991; Urbancic and541

Young , 1993; McGarr et al., 2010; Abercrombie, 1995; Abercrombie and Rice, 2005;542

Mayeda and Walter , 1996]. We fit the combined immature and mature populations543

using a linear regression for each population. The combined data set preserves the544

general feature of more self-similar behavior exhibited by the immature data sets in545

comparison to the size dependence exhibited by the mature data set. In particular,546

Figure 8 suggests that the earthquakes analyzed here follow the same trend as other547

populations of mature and immature events elsewhere on more ”typical” faults in548

southern California and the western United States.549

The seismicity induced in mines occurs as a direct result of the excavation ac-550

tivity, rather than long-term fault movement driven by tectonic stresses. Induced551

earthquakes can be assumed to occur on relatively new fracture surfaces, with low552

cumulative fault slip. The induced events may therefore be classified as immature for553

the purposes of comparison. The events included in the immature population shown554

in Figure 8 include seismicity recorded in the TauTona, and Mponeng gold mines in555

26

Page 27: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

South Africa [Kwiatek et al., 2011; McGarr et al., 2010; Yamada et al., 2007], and556

seismicity recorded in mines in Manitoba, and Ontario, Canada [Gibowicz et al., 1991;557

Urbancic and Young , 1993]. An additional advantage of using mining seismicity for558

a comparison of small earthquakes, is that the data quality is high, which is partic-559

ularly important for energy calculations. In all cases, earthquakes were recorded on560

instruments located at depths ranging from 50 - 3500 m, at source receiver distances561

on the order of ∼ 100 m. We apply the finite bandwidth correction given by Equation562

5 of Ide and Beroza [2001] to the data provided by Gibowicz et al. [1991]. All other563

data is taken directly as reported.564

By similar reasoning, the events analyzed in the preferred model (2) by Abercrom-565

bie [1995], and those re-analyzed in Abercrombie and Rice [2005] are analogous to566

the mature population, as they occur near the Cajon Pass section of the San An-567

dreas fault, which has a cumulative displacement similar to Parkfield. The results568

of Abercrombie [1995] use a Brune source model with a seismic QS = QP = 1000 to569

estimate the source parameters, rather than a spectral ratio approach. We assume570

that the comparison of our results is justified due to the high-quality borehole seismic571

data used, and the consistency of their results with the later analysis of Abercrombie572

and Rice [2005], which uses a spectral ratio approach similar to the one used here.573

Mayeda and Walter [1996] also use spectral ratio approach applied to earthquake574

coda spectra to determine the source function of a variety of events in the western575

US. The data include nearly all M > 5.0 earthquakes and selected aftershocks be-576

tween January 1988 and December 1994 (117 events total). They apply an empirical577

Green’s function correction to the coda spectra to obtain the source time functions,578

and calculate the seismic energy by integrating the moment rate spectra, with an579

27

Page 28: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

extrapolation assuming an ω−2 falloff. We omit the Northridge aftershocks included580

in their data set in the comparison (seven events total). The mainshock occurred on581

a previously unknown fault, and as a result, the maturity of the faults on which the582

aftershocks occur remains ambiguous. The remaining events occur on well-developed583

faults with surface traces of ∼10 km or more in the San Andreas, and Basin and584

Range faulting systems.585

We fit the composite dataset in Figure 8 and recover a similar trends to those586

in Figure 7. A linear regression indicates that the energy is roughly scale invariant587

for the combined immature population (slope of 0.07 ± 0.009), while the energy size588

dependence is more pronounced for the mature population (slope of 0.3 ± 0.01).589

The similarities in scaling between our observations and the combined data suggest590

that energy-size dependence scales similarly for events in Parkfield and the western591

US, as well as between the secondary faults in Parkfield, Mount St. Helens, and592

mining related events occurring on other immature faults. The only feature needed593

to distinguish the mature and immature populations in the comparison with other594

studies is the cumulative fault displacement. We propose that the similar scaling595

exhibited by the broader comparison with other studies suggests that the observed596

scaling features of the immature and mature populations presented here may be a597

general feature of earthquakes outside of our study areas, and not the result of some598

unusual faulting environment in the creeping/locked transition of the San Andreas599

fault, or in the volcanic edifice.600

28

Page 29: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

5 Implications for the Role of Fault Maturity in601

Earthquake Dynamics602

The comparison of energy-moment ratios (ER

M0,EG

M0) between immature and mature603

populations suggests that fault maturity affects the dynamics of rupture. More specif-604

ically, the energy relations in Figures 7 and 8 suggest that at least for smaller earth-605

quakes, less energy is radiated away, and less energy goes into forming new fractures606

on smoother, more mature faults compared to rougher, less mature faults. Other607

studies which compare the radiated energy of earthquakes between interplate events608

(inferred as mature) with intraplate events (inferred as immature) conclude that in-609

traplate earthquakes tend to radiate more energy that their interplate counterparts610

[Zhuo and Kanamori , 1987; Choy and Kirby , 2004; Choy and Boatwright , 2009]. The611

key difference in our study is that we differentiate immature and mature populations612

based on the amount of cumulative fault slip, rather than on focal mechanism or613

faulting style (e.g. strike slip, vs. thrust, or interplate vs. intraplate). The advantage614

of the approach is that it systematically compares the effect of maturity based on one615

fundamental physical parameter (displacement), which is more directly correlated616

with it.617

Harrington and Brodsky [2009] interpret the constant corner frequency scaling618

for the mature population as indicating that mature faults have source areas that619

change only nominally over the magnitude range of our data set. We consider the620

possible relationship between an unchanging source dimension and the decreasing621

energy-moment ratios observed here. The roughly constant fault area may be dictated622

by the asperity size, meaning that the smallest patch of the mature fault surface623

29

Page 30: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

that may rupture in an earthquake may be limited by the elliptical bump size. The624

scale invariant roughness observed for immature faults would imply that earthquake625

rupture might result from asperity rupture at a variety of length scales. In such626

a scenario, one would expect to observe a population of earthquakes with a fractal627

distribution of rupture areas, leading to an observation of self-similarity, or invariant628

energy-moment ratios (ER

M0, EG

M0). In fact, we observe a nearly constant energy values629

for the earthquakes in the immature population (Figure 7). We speculate that in a630

large event where the rupture length would exceed the elliptical bump (i.e. asperity)631

dimension by a significant amount, that fault slip and length may scale according to632

constant stress drop.633

The development of roughly constant source dimensions on the mature fault may634

result from a minimum asperity size asymptotically tending toward a fixed value.635

Observations of fault damage zones may also suggest that asperities trend toward a636

fixed size. Studies suggest that when a fault is first formed, the width of the damage637

zone increases with fault displacement. Such an effect occurs up to a certain offset,638

after which the rate of damage widening decreases [Savage and Brodsky , 2011; Chester639

and Chester , 2005]. Savage and Brodsky [2011] argue using observations of fracture640

density that the amount of displacement at which the damage rate decreases is ∼100641

m. Once cumulative displacement exceeds ∼ 100 m, the fault damage zone width is642

no longer determined by larger asperities, and the observed damage formation rate643

becomes roughly constant (see Savage and Brodsky [2011], Figure 4). They interpret644

the decline in damage formation dependence on asperity size as the result of slip645

being distributed throughout a wider patch on the fault surface, making it easier646

for the slipping patches to get around large asperities without damaging host rock.647

30

Page 31: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Consequently, earthquakes on less mature faults which accommodate more slip past648

asperities should have higher radiated energy values compared to their counterparts649

on mature fault surfaces. We claim that seismologically observable fracture energy650

values should be higher as well, if more asperities are being damaged. Figures 7651

and 8 indicate a higher ER, and EG per unit M0 for the immature population, and652

are consistent with such claims.653

6 Conclusions654

A comparison of ER

M0, and EG

M0between earthquakes on immature fault surfaces in655

Parkfield, California, and Mount St. Helens Volcano, Washington, and on a mature656

fault surface, namely the San Andreas fault near Parkfield, CA, indicates that the657

values are roughly constant for immature faults, while the values decrease with de-658

creasing magnitude for mature fault surfaces. Theoretical considerations based on659

a slip-weakening rupture model indicate a M0f3c dependence of static stress drop660

and energy ratios, which should remain constant for populations that exhibit self-661

similarity. The decreasing energy ratios and stress drop values with size observed662

for the mature population suggest that rupture properties are not self-similar for the663

mature population.664

The difference in source parameter scaling between immature and mature earth-665

quake populations suggests that the ordinary scaling relationships resulting from a666

commonly assumed constant stress drop are not generally valid for each earthquake667

population individually. The correlation of self-similar source parameter scaling with668

fault immaturity suggests that fault surface roughness may be a physical factor af-669

31

Page 32: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

fecting fault rupture.670

7 Acknowledgments671

The Mount St. Helens seismic data used here were collected by the Cascades Vol-672

cano Observatory, and the Pacific Northwest Seismograph Network, and distributed673

by the Incorporated Research Institutions for Seismology (IRIS). The IRIS DMS is674

funded through the National Science Foundation and specifically the GEO Direc-675

torate through the Instrumentation and Facilities Program of the National Science676

Foundation under Cooperative Agreement EAR-0552316. Data on fault locations and677

displacements was collected from the U.S. Geological Survey, California Geological678

Survey, 2006, Quaternary fault and fold database for the United States, accessed679

Sept. 15, 2009, from USGS web site: http//earthquakes.usgs.gov/regional/qfaults/.680

Parkfield waveform data is provided by Berkeley Seismological Laboratory, Uni-681

versity of California, Berkeley and accessible through the Northern California Earth-682

quake Data Center (NCEDC) website. The projected Landweber deconvolution code683

was provided by Hiroo Kanamori, and the FDDECON deconvolution code was pro-684

vided by Thorne Lay. This work was supported by the National Science Foundation685

grant EAR-0711575, and the Alexander Von Humboldt foundation.686

32

Page 33: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

References687

Abercrombie, R. E., Earthquake source scaling relationships from -1 to 5 M(L) using688

seismograms recorded at 2.5-km depth, J. Geophys. Res., 100, 24,015–24,036, 1995.689

Abercrombie, R. E., and J. R. Rice, Can observations of earthquake scaling constrain690

slip weakening?, Geophys. J. Int., 162, 406–424, 2005.691

Aki, K., and P. G. Richards, Quantitative Seismology, 2nd ed., Univ. Sci., Sausalito,692

CA, 2002.693

Allmann, B. P., and P. M. Shearer, Spatial and temporal stress drop variations in694

small earthquakes near Parkfield, California, J. Geophys. Res., 112, B04,305, 2007.695

Allmann, B. P., and P. M. Shearer, Global variations of stress drop for moderate to696

large earthquakes, J. Geophys. Res., 114 (B01310), 2009.697

Anderson, T. L., Fracture Mechanics: fundamentals and applications, 621 pp., CRC698

Press, 2004.699

Boatwright, J., and J. B. Fletcher, The partition of radiated energy between P and700

S waves, Bull. Seismol. Soc. Am., 74 (2), 1984.701

Brune, J. N., Tectonic stress and the spectra of seismic shear waves from earthquakes,702

J. Geophys. Res., 75, 4997–5009, 1970.703

Chester, F. M., and J. S. Chester, Fracture surface energy of the Punchbowl fault,704

San Andreas system, Nature, 437, 133–136, 2005.705

Choy, G. L., and J. Boatwright, Differential energy radiation from two earthquakes706

in Japan with identical M-w: the Kyushu 1996 and Tottori 2000 earthquakes, Bull.707

Seismol. Soc. Am., 99, 1815–1826, 2009.708

Choy, G. L., and S. H. Kirby, Apparent stress, fault maturity and seismic hazard for709

normal-fault earthquakes at subduction zones, Geophys. J. Int., 159, 2004.710

Choy, G. L., A. McGarr, S. H. Kirby, and J. Boatwright, An overview of the global711

variability in radiated energy and apparent stress, Geophysical Monograph, 170,712

43–57, 2006.713

Freund, L. B., The mechanics of dynamic shear crack propagation, J. Geophys. Res.,714

84 (B5), 2199–2209, 1979.715

33

Page 34: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Gibowicz, S. J., R. P. Young, S. Talebi, and D. J. Rawlence, Source parameters716

of seismic events at the underground research laboratory in Manitoba, Canada:717

Scaling relations for events with moment magnitude smaller than -2, Bull. Seismol.718

Soc. Am., 81 (4), 1157–1182, 1991.719

Hanks, T. C., and H. Kanamori, Moment magnitude scale, J. Geophys. Res., 84,720

2348–2350, 1979.721

Harrington, R. M., and E. E. Brodsky, Volcanic hybrid earthquakes that are brittle-722

failure events, Geophys. Res. Lett., 34 (L06308), 2007.723

Harrington, R. M., and E. E. Brodsky, Source duration scales with magnitude differ-724

ently for earthquakes on the San Andreas Fault and on secondary faults in Parkfield,725

CA, Bull. Seismol. Soc. Am., 99, 2323–2334, 2009.726

Ide, S., and G. C. Beroza, Does apparent stress vary with earthquake size?, Geophys.727

Res. Lett., 28 (17), 3349–3352, 2001.728

Ide, S., G. C. Beroza, S. G. Prejean, and W. L. Ellsworth, Apparent break in earth-729

quake scaling due to path and site effects on deep borehole recordings, J. Geophys.730

Res., 108 (2271), 2271, 2003.731

Imanishi, K., and W. L. Ellsworth, Source scaling relationships of microearthquakes732

at Parkfield, CA, determined using the SAFOD pilot hole seismic array, in Earth-733

quakes: Radiated Energy and the Physics of Faulting, edited by R. E. Abercrombie,734

A. McGarr, H. Kanamori, and G. Di Toro, Geophysical Monograph Series, pp. 81–735

90, AGU, 2006.736

Kanamori, H., and D. L. Anderson, Theoretical basis of some empirical relations in737

seismology, Bull. Seismol. Soc. Am., 65, 1073–1095, 1975.738

Kanamori, H., and E. E. Brodsky, The physics of earthquakes, Reports On Progress739

In Physics, 67, 1429–1496, 2004.740

Kanamori, H., and T. H. Heaton, American Geophysical Union Geophysical Mono-741

graph 120, chap. Microscopic and macroscopic physics of earthquakes, American742

Geophysical Union, 2000.743

Kanamori, H., and L. Rivera, Static and dynamic scaling relations for earthquakes744

and their implications for rupture speed and stress drop, Bull. Seismol. Soc. Am.,745

94, 314–319, 2004.746

34

Page 35: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Kanamori, H., J. Mori, E. Hauksson, T. Heaton, L. K. Hutton, and L. M. Jones, De-747

termination of earthquake energy-release and M(L) using Terrascope, Bull. Seismol.748

Soc. Am., 83, 330–346, 1993.749

Keilis-Borok, V., On estimation of the displacement in an earthquake source and of750

source dimensions, Ann. Geofis. (Rome), 12, 205–214, 1959.751

Kwiatek, G., K. Plenkers, and G. Dresen, Source parameters of picoseismicity752

recorded at Mponeng deep gold mine, South Africa: Implications for scaling re-753

lations, submitted to BSSA, 2011.754

Lanza, V., D. Spallarossa, M. Cattaneo, D. Bindi, and P. Augliera, Source parameters755

of small events using constrained deconvolution with empirical Green’s functions,756

Geophys. J. Int., 137, 651–662, 1999.757

Lay, T., and T. Wallace, Modern Global Seismology, 517 pp., Academic Press, San758

Diego, 1995.759

Lin, J., and R. Stein, Seismic constraints and Coulomb stress changes of a blind760

thrust fault system, 1: Coalinga and Kettleman Hills, California, USGS Open File761

Report, (1149), 17, 2006.762

Malagnini, L., S. Nielsen, K. Mayeda, and E. Boschi, Energy radiation from763

intermediate- to large-magnitude earthquakes: implications for dynamic fault weak-764

ening, J. Geophys. Res., 115 (B06319), 2010.765

Mayeda, K., and W. R. Walter, Moment, energy, stress drop, and source spectra of766

western United States earthquakes from regional coda envelopes, J. Geophys. Res.,767

101 (B5), 11,195–11,208, 1996.768

Mayeda, K., L. Malagnini, and W. R. Walter, A new spectral ratio method using769

narrow band coda envelopes: Evidence for non-self-similarity in the Hector Mine770

sequence, Geophys. Res. Lett., 34 (11), 2007.771

McGarr, A., J. B. Fletcher, M. Boettcher, N. Beeler, and J. Boatwright, Laboratory-772

based maximum slip rates in earthquake rupture zones and radiated energy, Bull.773

Seismol. Soc. Am., 100 (6), 3250–3260, 2010.774

Moran, S. C., S. D. Malone, A. I. Qamar, W. A. Thelen, A. K. Wright, and J. Caplan-775

Auerbach, Seismicity associated with renewed dome building at Mount St. Helens,776

20042005, USGS Open File Report, A volcano rekindled: the renewed eruption of777

Mount St. Helens, 2004-2006, (1750), 27–60, 2008.778

35

Page 36: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Ohnaka, M., A constitutive scaling law and a unified comprehension for frictional slip779

failure, shear fracture of intact rock, and earthquake rupture, J. Geophys. Res.,780

161 (9-10), 2003.781

Park, J., C. R. Lindberg, and F. L. Vernon III, Multitaper spectral analysis of high-782

frequency seismograms, J. Geophys. Res., 92, 12,675–12,684, 1987.783

Prejean, S. G., and W. L. Ellsworth, Observations of earthquake source parameters784

at 2 km depth in the Long Valley caldera, eastern California, Bull. Seismol. Soc.785

Am., 91, 165–177, 2001.786

Prieto, G. A., P. M. Shearer, F. L. Vernon, and D. Kilb, Earthquake source scaling787

and self-similarity estimation from stacking P and S spectra, J. Geophys. Res., 109,788

2004.789

Prieto, G. A., D. J. Thomson, F. L. Vernon, P. M. Shearer, and R. L. Parker, Confi-790

dence intervals for earthquake source parameters, Geophys. J. Int., 168, 1227–1234,791

2007.792

Revenaugh, J., and C. Reasoner, Cumulative offset of the san andreas fault in central793

California, Geology, 25 (2), 123–126, 1997.794

Sagy, A., and E. E. Brodsky, Geometric and rheological asperities in an exposed fault795

zone, J. Geophys. Res., 114 (B02301), 2009.796

Sagy, A., E. E. Brodsky, and G. J. Axen, Evolution of fault-surface roughness with797

slip, Geology, 35 (3), 283–286, 2007.798

Savage, H. M., and E. E. Brodsky, Collateral damage: Evolution with displacement of799

fracture distribution and secondary fault strands in fault damage zones, J. Geophys.800

Res., 116 (B3), B03,405, 2011.801

Schleicher, A. M., B. A. van der Pluijm, and L. N. Warr, Nanocoatings of clay and802

creep of the San Andreas fault at Parkfield, California, Geology, 38 (7), 667–670,803

2010.804

Shearer, P. M., G. A. Prieto, and E. Hauksson, Comprehensive analysis of earthquake805

source spectra in southern California, J. Geophys. Res., 111 (B06303), 2006.806

Simpson, R. W., M. Barall, J. Langbein, J. R. Murray, and M. J. Rymer, San Andreas807

fault geometry in the Parkfield, California, region, Bull. Seismol. Soc. Am., 96,808

S28–S37, 2006.809

36

Page 37: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Thurber, C., H. Zhang, F. Waldhauser, J. L. Hardebeck, A. Michael, and D. Eberhart-810

Phillips, Three-dimensional compressional wavespeed model, earthquake reloca-811

tions, and focal mechanisms for the Parkfield, California, region, Bull. Seismol.812

Soc. Am., 96, S38–S49, 2006.813

Tinti, E., P. Spudich, and M. Cocco, Earthquake fracture energy inferred from kine-814

matic rupture models on extended faults, J. Geophys. Res., 110 (B12), B12,303,815

2005.816

Titus, S. J., C. DeMets, and B. Tikoff, New slip rate estimates for the creeping817

segment of the San Andreas fault, California, Geology, 33, 205–208, 2005.818

Urbancic, T. I., and R. P. Young, Space-time variation in source parameters of mining-819

induced seismi events with M ¡ 0, Bull. Seismol. Soc. Am., 83 (2), 378–397, 1993.820

Venkataraman, A., and H. Kanamori, Observational constraints on the fracture en-821

ergy of subduction zone earthquakes, J. Geophys. Res., 109 (B05302), 2004a.822

Venkataraman, A., and H. Kanamori, Effect of directivity on estimates of radiated823

seismic energy, J. Geophys. Res., 109 (B04301), 2004b.824

Venkataraman, A., L. Rivera, and H. Kanamori, Radiated energy from the 16 October825

1999 Hector Mine earthquake: Regional and teleseismic estimates, Bull. Seismol.826

Soc. Am., 92, 1256–1265, 2002.827

Venkataraman, A., G. C. Beroza, and J. Boatwright, A brief review of techniques used828

to estimate radiated seismic energy, Geophysical Monograph, 170, 15–24, 2006a.829

Venkataraman, A., G. C. Beroza, S. Ide, K. Imanishi, H. Ito, and Y. Iio, Measurements830

of spectral similarity for microearthquakes in western Nagano, Japan, J. Geophys.831

Res., 111, 2006b.832

Waldhauser, F., and W. L. Ellsworth, A double-difference earthquake location algo-833

rithm: Method and application to the northern Hayward fault, California, Bull.834

Seismol. Soc. Am., 90, 1353–1368, 2000.835

Waldhauser, F., and W. L. Ellsworth, Fault structure and mechanics of the Hayward836

fault, California, from double-difference earthquake locations, J. Geophys. Res.,837

107 (2054), 2002.838

Wesnousky, S. G., Seismological and structural evolution of strike-slip faults, 1988.839

Wesnousky, S. G., Predicting the endpoints of earthquake ruptures, Nature, 444,840

358–360, doi:10.1038, 2006.841

37

Page 38: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Wyss, M., and J. N. Brune, Seismic moment, stress, and source dimentions for earth-842

quakes in the california-nevada region, J. Geophys. Res., 73, 4681, 1968.843

Yamada, T., J. Mori, S. Ide, R. E. Abercrombie, H. Kawakata, M. Nakatani, Y. Iio,844

and H. Ogasawara, Stress drops and radiate seismic energies of microearthquakes845

in a South African gold mine, J. Geophys. Res., 112 (B03305), 2007.846

Zhuo, Y., and H. Kanamori, Regional variation of the short-period (1 to 10 second)847

source spectrum, Bull. Seismol. Soc. Am., 77, 514–529, 1987.848

38

Page 39: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

A Analytical expression for radiated energy849

Following Ide and Beroza [2001], we re-derive the analytical expression for ER here850

leading to the expression for radiated energy given in Equation 5. Equation 1 of Ide851

& Beroza, 2001 is the expression for a velocity spectrum based on a Brune spectral852

model:853

ˆu(f) =M0

4πρα3r

f

[1 + (f/fc)2]=

Ω0f

[1 + (f/fc)2](A.1)854

with an assumed spectral falloff of n = 2. Note that the integrated spectral velocity855

in Equation 3 differs from that given in Ide and Beroza [2001] in that it contains the856

spectral amplitude measured at the station (Ω0) instead of M0. In order to obtain857

the seismic moment, the observable Ω0 must be multiplied by the source-receiver858

distance (r), as shown in Equation A.1. I in Equation 3 is therefore proportional to859

the radiated energy. The constant of proportionality is given by Equation 4.860

Calculating the integrated velocity I using the power spectral density of the veloc-861

ity spectrum measured at a single station as given by the expression in Equation 3,862

and assuming an ideal case of an unlimited bandwidth (i.e. integrating from 0 to863

infinite frequency),864

I =

∫ ∞−∞

ˆu2(f)df = 8π2Ω20

∫ ∞0

f 2

[1 + (f/fc)2]2df (A.2)865

I = 8π2Ω20f

4c

[arctan(f/fc)

2fc− f

2(f 2c + f 2)

]∣∣∣∣∣∞

0

. (A.3)866

If we define F (f, fc) as follows,867

39

Page 40: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

F (f, fc) ≡−f/fc

1 + (f/fc)2+ arctan(f/fc), (A.4)868

then869

I = 8π2Ω20f

3c F (f, fc)

∣∣∣∣∣∞

0

= 2π2Ω20f

3c (π

2+ 0), (A.5)870

I = 2π3Ω20f

3c . (A.6)871

Substituting Equation A.6 into Equation 4 gives us:872

ER = 〈U2θφ〉R2

pψ8π4ρβr2Ω20f

3c . (A.7)873

Substituting the expression in Equation 2 relating Ω0 and r with M0 leads to the874

analytical expression for radiated energy in terms of moment and corner frequency875

given by Equation 5:876

ER =π2

5ρβ5M2

0 f3c . (A.8)877

40

Page 41: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Date Time, UTC Lat. Lon. Depth Magnitude Event ID fc (Hz)2004/09/05 07:05:04.86 35.7685 -120.319 8.369 1.5 21393520 14.42004/09/28 17:57:43.98 35.7784 -120.33 8.388 2.5 21400463 15.62004/09/28 19:40:23.6 35.7785 -120.33 8.649 2.6 21400518 16.12004/09/28 20:55:07.64 35.7818 -120.323 4.161 2.0 51148134 15.02004/09/28 21:49:23.56 35.7771 -120.329 8.575 2.7 21400488 7.42004/09/28 21:57:37.88 35.7983 -120.342 4.163 2.0 51148203 15.02004/09/29 18:59:59.92 35.7827 -120.334 8.328 1.8 51148847 15.22004/09/30 01:57:50.06 35.7812 -120.323 3.97 1.6 21400936 16.22004/10/07 14:32:03.64 35.781 -120.332 8.593 2.2 51150206 12.02004/10/18 06:22:57.07 35.7826 -120.334 8.192 1.8 51151460 14.92004/10/18 06:27:51.91 35.7822 -120.334 8.279 1.6 51151464 14.52004/10/19 15:52:17.48 35.7699 -120.321 8.619 1.9 51151568 14.62004/10/20 09:09:47.86 35.7978 -120.341 4.268 2.2 51151627 14.72004/10/29 03:32:43.73 35.781 -120.333 8.751 3.0 21415362 13.42004/11/03 14:42:54.09 35.7667 -120.318 8.324 1.6 21416910 4.52001/09/20 20:06:02.64 35.9347 -120.487 5.261 2.1 21194856 20.92004/09/28 18:37:16.76 35.9345 -120.487 5.216 1.6 51148006 19.32004/09/28 23:33:49.64 35.9347 -120.487 5.272 2.2 21400519 18.32004/10/02 23:25:08.72 35.9354 -120.487 5.349 2.3 51149613 18.52005/01/08 04:46:6.00 35.9346 -120.487 5.27 2.1 21432539 18.22004/02/04 14:29:57.39 36.0951 -120.66 7.518 1.2 21340072 24.82005/02/09 10:45:23.62 36.095 -120.66 7.518 1.6 21438412 20.42005/02/09 10:53:15.6 36.0953 -120.66 7.527 1.2 21438414 16.82006/03/11 08:25:42.98 36.0362 -120.596 4.82 2.2 21508986 13.52007/01/17 10:11:35.6 36.037 -120.595 4.83 2.1 51177776 17.92002/06/04 22:33:56.28 35.932 -120.676 10.163 2.1 21228776 23.52002/06/04 22:50:02.08 35.932 -120.676 10.168 2.1 21228784 16.62004/09/06 03:43:04.43 36.148 -120.653 4.398 2.6 21393628 6.12004/09/26 15:54:05.96 36.143 -120.666 4.48 3.2 21399972 4.82004/09/27 08:52:40.96 36.154 -120.658 4.821 2.5 21400160 9.72004/09/27 10:38:56.32 36.152 -120.658 4.32 2.0 21400192 12.22004/10/03 02:11:15.36 36.153 -120.658 4.347 1.5 21402827 13.72006/06/27 21:38:11.87 36.065 -120.192 16.95 3.9 21524551 8.72006/12/15 19:50:25.33 36.170 -120.298 9.96 3.2 51176831 9.72007/03/12 12:13:54.63 35.938 -120.691 8.89 1.4 40194471 31.62007/09/20 05:41:07.00 36.064 -120.194 14.34 2.0 40202213 8.8

Table 1: Parkfield earthquakes considered in this study, taken from the catalog ofThurber et al. [2006], and from the ANSS catalog. (Events in Mount St. Helens arenot cataloged). The 25 events located on the San Andreas Fault are grouped intothe three clusters from southeast to northwest (Figure 1). Each cluster is separatedby a single line. The 11 events located on secondary faults (bottom) are separatedby a double horizontal line. Duration is taken as the width of the source time pulse[Harrington and Brodsky , 2009], and corner frequencies are determined using thesource time function and a least squares fit to a Brune spectral model. Magnitudevalues are calculated according to the method described in the text.

41

Page 42: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

moment, M0 (Nm) corner frequency, fc (Hz)2.8e10 8.31.4e11 8.13.6e10 9.14.7e10 13.34.6e10 11.62.8e10 9.71.0e11 8.42.4e10 11.95.6e10 9.34.3e10 10.91.2e11 5.13.6e10 9.63.1e10 9.66.6e10 13.36.3e10 10.08.2e10 6.74.5e10 9.94.7e10 9.45.0e9 12.15.4e10 8.94.7e10 13.02.7e10 15.04.6e10 13.04.0e10 12.86.0e10 7.95.0e10 12.24.1e10 8.59.7e10 10.03.7e10 12.89.8e10 10.39.8e10 8.71.1e11 9.72.2e10 13.14.7e10 11.42.8e10 11.24.6e10 11.43.8e10 11.55.5e10 10.94.7e10 11.05.5e10 9.8

Table 2: Seismic moment and corner frequency values of Mount St. Helens earth-quakes calculated here. Events are not cataloged, and therefore have no origin timeor epicentral location. Empirical Green’s function moment value calculated via thelong-period spectral amplitude, and remaining relative moment and corner frequencyvalues originate from a least-squares fit to a Brune spectral model of the source timefunction (see text). Source time functions determined with a water-level spectraldeconvolution.

42

Page 43: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

0 5 10 km

120.7° W 120.6° W 120.5° W 120.4° W 120.3° W 120.2° W

35.8° N

35.9° N

36.0° N

36.1° N

36.2° N

mature

immature

1966 and 2004 mainshocks

HRSN Stations

Figure 1: Parkfield study area. The map shows the location of the mature (circles),and immature (diamonds) events used in our study. Symbols are scaled accordingto earthquake duration as determined from the source time function pulse width.Source time functions for these earthquakes are obtained by projected Landweberdeconvolution of a co-located earthquake that is at least one magnitude unit smallerthan the events shown. The large stars represent the 1966 and 2004 earthquakeepicenters. Symbol shapes and colors follow the same convention in the subsequentfigures.

43

Page 44: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

Figure 2: Mount St. Helens study area. Grayscale indicates elevation in feet.The map shows the location of cataloged earthquakes occurring from 2/26/2005 to3/8/2005 (the time of our data set). Note that events shown here are meant to give anindication of the location of cataloged seismicity during the time period of our datacollection. They do not correspond to the exact location of the events in our data set.(Earthquakes analyzed here are below the catalog threshold). S-P wave arrival times,as well as a lack of seismicity elsewhere in the volcanic edifice indicate that theirlocations are in the crater. Stations MIDE and NED of the Cascades Chain networkare shown as white triangles. Additional stations used for S-wave velocity estimationindicated by white squares (the 7th additional station used, JUN, is located to thenorth, outside the map boundary, at 45.15 N, 122.22 W).

44

Page 45: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

−101

x 105

VCAB vertical

Main Event

−101

x 105

VCAB horizontal

0 5 10

−101

x 105

VCAB horizontal

−10000

1000

eGf

−10000

1000

0 5 10

−10000

1000

0 2 4 6−500

0

500MSH

MIDE vertical

time (s)0 2 4

−20

0

20MSH

vert

ical

time (s)

Figure 3: Examples of two earthquake-empirical Green’s functions (eGf) pairs fromstation VCAB in Parkfield (three components shown in the top three rows) andstation MIDE at Mount St. Helens (bottom row). eGf’s shown in the right column,and events for which the source-time function is calculated shown in left column. Themain event and eGF Mw values respectively for Parkfield are 2.1, and 0.8, and forMount St. Helens are 1.2, and 0.07.

45

Page 46: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

0 1 2 3 4

0

2000

4000VCAB vertical

0 1 2 3 4

0

2000

4000VCAB horizontal

0 1 2 3 4

0

2000

4000VCAB horizontal

Am

plitu

de R

atio

0 1 2 3 4

0

1000

2000MIDE vertical

time (s)

100

101

102

100

101

102

103

spec

tral

am

p. r

atio

VCAB

100

101

100

101

102

103

spec

tral

am

p. r

atio

frequency (Hz)

MIDE

Figure 4: Left: source-time functions calculated using the event pairs shown in Fig-ure 3. Three components for station VCAB shown in top three rows. The source-timefunctions in Parkfield determined using a Projected Landweber deconvolution, andthose at Mount St. Helens determined using a water level deconvolution. Water leveldeconvolution in Parkfield depicted as light gray line in top three panels on the leftto show consistency with the Projected Landweber deconvolution (see text). Source-time function from MIDE vertical component shown on the bottom. Right: Sourcespectra calculated from the three-component, RMS-spectra at station VCAB (top)and MIDE bottom. The least-squares fit to a Brune spectra shown with dashed line,and the noise spectra shown with a thin, gray line.

46

Page 47: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

0.5 1 1.5 2 2.5 3 3.5 4 4.50.5

1

1.5

2

2.5

3

3.5

4

4.5

Mc (catalog)

Mw

(ca

lcul

ated

)

matureimmatureimmature (extra stations)1:1

Figure 5: Moment magnitude values (MW ) calculated in this study compared withcatalog magnitude values (predominantly duration magnitudes Md) [Thurber et al.,2006]. The average ratio of Mw : Mc is 0.92±0.04 for the mature (blue) events,and 1.05±0.13 for the immature events (red diamonds). The scatter in immatureevents likely results from a lack of azimuthal coverage from HRSN stations (Figure 1).The red squares indicate Mw calculations based on moment values estimated usingadditional stations (see text).

47

Page 48: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

109

1010

1011

1012

1013

1014

1015

100

101

102

moment (Nm)

f c (H

z)

constant ∆ σM

0 ∝ f

c−0.03

Figure 6: Seismic moment vs. corner frequency estimated using a least-squared fit toa Brune spectral model. Red diamonds and squares indicate the immature populationin Mount St. Helens and Parkfield respectively. Blue circles represent the maturepopulation in Parkfield. Filled symbols indicate seismic moments estimated here usinglong-period spectral ratios, and open symbols indicate moment values calculated usingconverted catalog values for comparison. Error bars not visible on filled symbols aresmaller than the symbol size. The dashed line indicates a least squares fit to themoment-corner frequency trend for the mature population, and the fit indicates thatfc is independent of M0 to within error for the mature population. The solid linerepresents lines of constant stress drop (∆σ = 2 MPa, β = 800 m/s for MSH, and∆σ = 10 MPa, β = 3750 m/s for Parkfield). Immature events in Parkfield and MSHfollow two separate lines of constant stress drop because the shear wave velocity atMSH is roughly four times slower than at Parkfield.

48

Page 49: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

1e−9

1e−8

1e−7

1e−6

1e−5

1e−4

1e−3

ER/M

0

109

1010

1011

1012

1013

1014

1015

1016

1e−8

1e−7

1e−6

1e−5

1e−4

1e−3

EG

/M0

immature (Parkfield)immature (MSH)mature (Parkfield)immature (catalog moment)mature (catalog moment)slope =0.9 ± 0.16slope = −0.2 ± 0.08

Figure 7: Ratio of radiated seismic energy vs. seismic moment (ER

M0, left axis) and

ratio of seismically observable fracture energy (EG

M0, right axis) vs. earthquake size.

Earthquakes on immature faults are represented by filled diamonds, and squares.Earthquakes on a mature fault are represented by filled circles. Error bars not visibleon filled symbols are smaller than the symbol size. Values of moment taken from thecatalog are shown for comparison as open symbols. A comparison between valuesfrom earthquakes on mature, vs. immature faults indicates that the ratio remainsroughly constant for earthquakes on immature, or rougher faults, (as expected forconstant stress drop scaling) and decreases with moment for earthquakes on mature,smoother faults.

49

Page 50: Energy and Rupture Dynamics are Di erent forseisweb/emily_brodsky/reprints/... · 2011. 5. 4. · 1 Energy and Rupture Dynamics are Di erent for 2 Earthquakes on Mature Faults vs.

100

102

104

106

108

1010

1012

1014

1016

1018

1020

1022

1e−10

1e−9

1e−8

1e−7

1e−6

1e−5

1e−4

1e−3

ER/M

0 (B

rune

mod

el)

Moment (Nm)

1e−9

1e−8

1e−7

1e−6

1e−5

1e−4

1e−3

EG

/M0immature (Parkfield)

immature (MSH)Yamada et al., 2005Kwiatek et al., submittedUbancic et al., 1993Gibowicz et al., 1991McGarr et al., 2010matureAbercrombie 1995 (Model 2)Abercrombie & Rice, 2005Mayeda and Walter, 1996slope = 0.3 ± 0.01slope = 0.07 ± 0.009

Figure 8: Figure 7 shown together with the data points [Yamada et al., 2007; Kwiateket al., 2011; Gibowicz et al., 1991; Urbancic and Young , 1993; McGarr et al., 2010;Abercrombie, 1995; Abercrombie and Rice, 2005; Mayeda and Walter , 1996] for com-parison, suggesting that our results are a general feature. Events comprised largely ofmining induced seismicity are representative of an immature population, while eventsin the western US (San Andreas and Basin and Range faulting systems) are repre-sentative of a mature population (see text). We estimate the size dependence of thecombined immature and mature populations using a linear regression. The estimatedslope values shown in the legend are for the combined data sets.

50


Recommended