Hector Mine Research Articles

Nonuniform prestress from prior earthquakes and the effect on dynamics of branched fault systems

To examine the effects of branched fault geometry on the dynamics of fault systems in the long term, we perform multicycle simulations on generic faulting models. An explicit finite element algorithm is used to simulate spontaneous dynamic rupture of earthquakes. The fault stress during the interseismic period is evaluated by an analytical viscoelastic model. We find that the fault prestress field becomes highly nonuniform near the branch point and on the two branch segments over multiple earthquake cycles, owing to the branched fault geometry and stress interaction between the two segments. The principal prestress on faults rotates over multiple earthquake cycles and departs from the regional stress field significantly near the branch point. After a number of earthquake cycles, the branched fault systems evolve to a steady state in which several patterns of the fault prestress and earthquake rupture repeat. The nonuniform prestress developed from previous earthquakes has large effects on the rupture and slip patterns. Several different rupture scenarios can occur on a given branched fault system. In addition, backward branching can occur in the nonuniform prestress field, either driven by slip on the “stem” of the fault system or through a triggering mechanism. These modeling results may have important implications for understanding fault‐branching behavior observed in the 1992 Landers, the 1999 Hector Mine, and the 2002 Denali fault earthquakes and for seismic hazard analysis in the areas where branched fault systems exist.

Bayesian Hierarchical Space–time Modeling of Earthquake Data

Stochastic earthquake models are often based on a marked point process approach as for instance presented in Vere-Jones (Int. J. Forecast., 11:503–538, 1995). This gives a fine resolution both in space and time making it possible to represent each earthquake. However, it is not obvious that this approach is advantageous when aiming at earthquake predictions. In the present paper we take a coarse point of view considering grid cells of 0.5 × 0.5°, or about 50 × 50 km, and time periods of 4 months, which seems suitable for predictions. More specifically, we will discuss different alternatives of a Bayesian hierarchical space–time model in the spirit of Wikle et al. (Environ. Ecol. Stat., 5:117–154, 1998). For each time period the observations are the magnitudes of the largest observed earthquake within each grid cell. As data we apply parts of an earthquake catalogue provided by The Northern California Earthquake Data Center where we limit ourselves to the area 32–37° N, 115–120° W for the time period January 1981 through December 1999 containing the Landers and Hector Mine earthquakes of magnitudes, respectively, 7.3 and 7.1 on the Richter scale. Based on space-time model alternatives one step earthquake predictions for the time periods containing these two events for all grid cells are arrived at. The model alternatives are implemented within an MCMC framework in Matlab. The model alternative that gives the overall best predictions based on a standard loss is claimed to give new knowledge on the spatial and time related dependencies between earthquakes. Also considering a specially designed loss using spatially averages of the 90th percentiles of the predicted values distribution of each cell it is clear that the best model predicts the high risk areas rather well. By using these percentiles we believe that one has a valuable tool for defining high and low risk areas in a region in short term predictions.

Long‐period earthquake ground displacements recorded on Guadeloupe (French Antilles)

AbstractDisplacement time‐histories derived from accelerograms of three recent earthquakes in western North America (Hector Mine, Mw 7.1; Denali, Mw 7.9; and San Simeon, Mw 6.5) have been shown to feature large long‐period (∼10 s) ground‐motion cycles. Such long‐period displacements cause a localized peak within the displacement response spectrum that is currently not considered within any earthquake engineering design spectra. These displacement pulses have also been shown to be persistent and to feature on time‐histories from widely separated stations (∼20 km).Broadband and accelerometric data from the Les Saintes earthquake sequence of 2004–2006 (4.9⩽Mw⩽5.3) recorded on Guadeloupe (French Antilles) are shown in this article to feature similar long‐period motions. The broadband data are used to independently corroborate the displacement time‐histories derived through high‐pass filtering and double integration of accelerometric data. It is shown that high‐quality broadband data are suitable for this purpose. The long‐period motions observed cause a localized peak in displacement response spectra at periods between 5 and 10 s. It is suggested here that the cause of these large‐amplitude long‐period motions are specific source mechanisms, which may possibly involve the presence of fluids within the source.The form of the displacement response spectra from these time‐histories is significantly different from the spectral shape specified in recent seismic design codes since the peak in the spectra is at a much greater period than expected. This leads to an underestimation of spectral displacements for periods between about 5 and 10 s. Therefore, if these observed long‐period cycles are a common feature of earthquake ground motions the standard form of displacement design spectra may need to be reconsidered. Copyright © 2007 John Wiley & Sons, Ltd.

Postearthquake Relaxation and Aftershock Accumulation Linearly Related after the 2003 M 6.5 Chengkung, Taiwan, and the 2004 M 6.0 Parkfield, California, Earthquakes

We treat both the number of earthquakes and the deformation following a mainshock as the superposition of a steady background accumulation and the postearthquake process. The preseismic displacement and seismicity rates r u and r E are used as estimates of the background rates. Let t be the time after the mainshock, u ( t ) + u 0 the postseismic displacement less the background accumulation r u t , and Δ N ( t ) the observed cumulative number of postseismic earthquakes less the background accumulation r E t . For the first 160 days (duration limited by the occurrence of another nearby earthquake) following the Chengkung ( M 6.5, 10 December 2003, eastern Taiwan) and the first 560 days following the Parkfield ( M 6.0, 28 September 2004, central California) earthquakes u ( t ) + u 0 is a linear function of Δ N ( t ). The aftershock accumulation Δ N ( t ) for both earthquakes is described by the modified Omori Law d Δ N / dt ∝ (1 + t / τ ) − p with p = 0.96 and τ = 0.03 days. Although the Chengkung earthquake involved sinistral, reverse slip on a moderately dipping fault and the Parkfield earthquake right-lateral slip on a near-vertical fault, the earthquakes share an unusual feature: both occurred on faults exhibiting interseismic fault creep at the surface. The source of the observed postseismic deformation appears to be afterslip on the coseismic rupture. The linear relation between u ( t ) + u 0 and N ( t ) suggests that this afterslip also generates the aftershocks. The linear relation between u ( t ) + u 0 and Δ N ( t ) obtains after neither the 1999 M 7.1 Hector Mine (southern California) nor the 1999 M 7.6 Chi-Chi (central Taiwan) earthquakes, neither of which occurred on fault segments exhibiting fault creep.

What Controls the Spatial Distribution of Remote Aftershocks?

Motivated by recent studies showing that stress modifications due to small quakes are important in earthquake triggering, we use a quasi-dynamic spatially discrete model of a rate- and state-dependent fault to examine the effect of multiple elastic stress transfers on the spatial extent of aftershock activity. We show that multiple stress transfers may significantly alter the spatial distribution of aftershock sequences, and give rise to stress-seismicity relations that are inconsistent with Dieterich’s aftershock model. Specifically, we present an example in which, owing to the effect of multiple elastic stress transfers, the area experiencing seismicity rate change is much larger than that subjected to a stress change. We define a parameter that quantifies the proximity to the failure in the model, and show that differences in the magnitude of aftershock rates are due to differences in the proximity to failure that prevailed prior to the mainshock. Furthermore, we show that accelerating and decelerating cumulative Benioff strains are indicative of a region approaching or moving away from failure, respectively. Finally, we compare the cumulative Benioff strains in remote sites triggered either by the Landers or the Hector Mine earthquake, and infer that differences in the state of stress that prevailed prior to these mainshocks may explain the different spatial distribution of their remote aftershocks.

Critical point theory of earthquakes: Observation of correlated and cooperative behavior on earthquake fault systems

The critical point theory for earthquakes was originally proposed to explain the scaling relations observed in earthquakes, including the Gutenberg‐Richter frequency‐ magnitude relation and the Omori's law for aftershocks. In this model, main shocks, their foreshocks and aftershocks are all associated with the formation of a correlated, cooperative spatial region with high stress. Until now, only indirect evidence of the existence of these correlated regions has been reported. Here in this paper we present observations and analyses that allow us to directly map the high stress, spatially correlated regions preceding four major earthquakes, i.e. the 1992 Landers (California), 1995 Kobe (Japan), 1999 Chi‐Chi (Taiwan) and 1999 Hector Mine (California) earthquakes. We therefore conclude that the locations and extent of large main shocks and their immediate aftershocks can be determined from seismicity data taken prior to the main shocks, and provide additional evidence in support of the critical point theory for earthquakes.

Measuring two‐dimensional movements using a single InSAR pair

We present a new method to extract along‐track displacements from InSAR data, based on split‐beam InSAR processing, to create forward‐ and backward‐looking interferograms. The phase difference between the two modified interferograms provides the along‐track displacement component. Thus, from each conventional InSAR pair we extract two components of the displacement vector: one along the line of sight, the other in the along‐track direction. We analyze the precision of the new method by comparing our solution to GPS and offset‐derived along‐track displacements in interferograms of the 1999, Hector Mine earthquake. The RMS error between GPS displacements and our results ranges from 5 to 8.8cm. Our method is consistent with along‐track displacements derived by pixel‐offsets, which are limited to 12–15cm precision. The theoretical precision of the new method depends on SNR and coherence. For a signal to noise ratio of 30 the expected precisions are 3, 11cm for coherence of 0.8, 0.4, respectively.

The Absence of Remotely Triggered Seismicity in Japan

Remote dynamic triggering occurred in the western United States and Greece following the following the Denali, Hector Mine, Landers, and Izmit main- shocks in seismically active, largely geothermal areas that experienced shaking from the mainshock typically exceeding 0.2 cm/sec. Because triggered earthquakes are small (often below Mw, 2), investigations of whether remote triggering is a generally occurring phenomena must be concentrated in well-instrumented locations. As a seismically active, geothermal, well-instrumented location, Japan constitutes an ideal location for such study. We systematically examine whether remote dynamic trig- gering occurs following mainshocks in Japan producing amplitudes in the study region above 0.2 cm/sec at distances greater than two estimated fault lengths, from 1998 to 2004. We use both catalog data and filtered waveforms to search for seis- micity increases. One significant regional increase follows a Mw 7.6 mainshock in March 2000, largely due to a volcanic swarm that began the day before and increased in intensity in the following days. Two localized increases in cataloged seismicity occur in Kyushu, which is the only area in the study region with onshore tectonic extension. Two factors possibly inhibit regional triggering in Japan: (1) compres- sional tectonics, and (2) the frequent occurrence of large mainshocks. The ambient seismicity in Japan combined with the lack of observable widespread triggering ren- ders rate-state friction and stress corrosion to be improbable triggering mechanisms. Fracture unclogging as a triggering mechanism is consistent with the observations.

Analysis of strain‐induced ground‐water fluctuations at Devils Hole, Nevada

AbstractThe pool in Devils Hole is a sensitive indicator of crustal strain and fluctuates in response to changes in atmospheric pressure, earth tides, earthquakes, large‐scale tectonic activity and ground‐water development. Short‐term and cyclic water‐level fluctuations caused by atmospheric pressure and earth tides were found to be on the order of millimeters to centimeters. The 1992 Landers/Little Skull Mountain earthquake sequence and the 1999 Hector Mine earthquake induced water‐level offsets of greater than −12 and −3.6 cm, respectively. The results of a dislocation model used to compute volumetric strain for each earthquake indicates that the coseismic water‐level offsets are consistent in magnitude and sense with poroelastic responses to earthquake‐induced strain. Theoretical postseismic fluid‐flow modeling indicates that the diffusivity of the system is on the order of 0.03 m2 sec−1, and identified areas of anomalous water‐level fluctuations. Interpretation of model results suggests that while the persistent post‐Landers rise in water‐level can be attributed to deformation‐induced channeling of fluid to the Devils Hole fault zone, the cause of the pre‐Hector Mine water‐level rise may be related to postseismic excess fluid pressures or preseismic strain accumulation.

On the Role of Multiple Interactions in Remote Aftershock Triggering: The Landers and the Hector Mine Case Studies

The observed seismicity rate increase after large earthquakes in sites that are located several source lengths away from the mainshock centroid poses a major problem. This is because the static stress change induced by a mainshock in that region seems to be insignificant, and the dynamic stress changes can only en- hance the seismicity during the passage of the seismic waves but not at later times. In quest for a physically viable triggering mechanism for delayed aftershocks in remote sites, we examine earthquake activities in remote sites that were triggered by two California earthquakes, the magnitude 7.1 Hector Mine earthquake and the mag- nitude 7.3 Landers earthquake. We introduce a new method for quantifying the de- gree to which the triggering effect of each aftershock is locally more important than the triggering effect of the mainshock. We apply this method to the Landers and the Hector Mine remote aftershock sequences. We show that multiple stress transfers from early aftershocks to later aftershocks played an important role in the enhance- ment of both the Landers and the Hector Mine aftershock activities in remote sites. We present a time-space diagram of the Hector Mine remote aftershock sequence in the Imperial Valley, which shows that this sequence is made up of several subse- quences and that the onset of activity migrated southward.

The physical nature of thermal anomalies observed before strong earthquakes

The paper examines the effect of air ionization on the thermal balance of the boundary layer of atmosphere. In seismically active areas the increased radon emanation from active faults and cracks before earthquakes is the primary source of air ionization. The problem is analyzed both on microscopic and macroscopic levels and in both cases the significant changes of the air relative humidity and air temperature are obtained. This happens due to the water molecules attachment to the newly formed ions (or in other words, condensation) which leads to the excretion of the latent heat. Obtained results permit us to explain the changes of the surface temperature and the surface latent heat flux increase before earthquakes observed by remote sensing satellites, as well as ground based measurements of the air temperature and relative humidity variations before the Colima earthquake (M7.6) of 2003 in Mexico, Hector Mine earthquake (M7.1) of 1999 in USA and Parkfield earthquake (M6) of 2004 in USA. These findings are also supported by the results of active experiments where the installation of artificial ionization of atmosphere is used.

Estimating Locations and Magnitudes of Earthquakes in Southern California from Modified Mercalli Intensities

Modified Mercalli intensity (mmi) assignments, instrumental moment magnitudes M , and epicenter locations of thirteen 5.6 ≤ M ≤ 7.1 “training-set” events in southern California were used to obtain the attenuation relation mmi = 1.64 + 1.41 M − 0.00526 * Δ h − 2.63 * log Δ h , where Δ h is the hypocentral distance in kilometers and M is moment magnitude. Intensity magnitudes M I and locations for five 5.9 ≤ M ≤ 7.3 independent test events were consistent with the instrumental source parameters. Fourteen “historical” earthquakes between 1890 and 1927 were then analyzed. Of particular interest are the M I 7.2 9 February 1890 and M I 6.6 28 May 1892 earthquakes, which were previously assumed to have occurred near the southern San Jacinto fault; a more likely location is in the Eastern California Shear Zone (ecsz). These events, and the 1992 M 7.3 Landers and 1999 M 7.1 Hector Mine events, suggest that the ecsz has been seismically active since at least the end of the nineteenth century. The earthquake catalog completeness level in the ecsz is ∼ M 6.5 at least until the early twentieth century. Online material : Figures and table of empirical mmi site corrections.

3D Cohesive End-Zone Model for Source Scaling of Strike-Slip Interplate Earthquakes

A 3D, static fracture mechanics model of earthquake rupture that in- corporates cohesive end zones (CEZs), or zones of increased frictional strength, is tested to determine whether it helps to understand the observed scaling behavior of average slip with rupture dimensions for shallow (! 20 km), continental, interplate strike-slip earthquakes. Our new compilation of average source parameters suggests that (1) average slip increases with aspect ratio (along-strike length/down-dip width), although in decreasing proportions for progressively larger aspect ratio ruptures, and (2) a gradual scaling change exists at an aspect ratio of 6. These general trends match the functional form predicted by the CEZ model. Despite these general trends, significant scatter in average slip is apparent among similarly sized ruptures. We test the hypothesis that the CEZs represent strength heterogeneities along the rupture surface that result from velocity-strengthening frictional behavior; and that this het- erogeneity in frictional behavior along the fault is the primary reason for the failure of a universal (constant stress drop) scaling law. CEZ lengths are measured from slip and stress drop distributions determined from published inversions of geophysical data for the 1984 Morgan Hill, 1992 Landers, 1999 Hector Mine, and 1999 I zmit earthquakes, and range from 15 to 40% of the rupture segment lengths. These lengths are an order of magnitude larger than that inferred from characteristics of high-frequency seismic radiation (i.e., f max ). These data indicate that the ratio of average coseismic slip to rupture length decreases in the presence of large CEZs. Measured CEZ lengths, rupture dimensions, and average slip are used to calculate average resolved shear-driving stresses and CEZ shear-yield strengths on the order of 10-30 MPa. In our new model of earthquake rupture, stress drop is predicted to be a small fraction of fault strength and thus supports a partial stress drop model of earthquake rupture for strike-slip interplate events.

Postseismic relaxation and transient creep

Postseismic deformation has been observed in the epicentral area following the 1992 Landers (M = 7.3), 1999 Chi‐Chi (M = 7.6), 1999 Hector Mine (M = 7.1), 2002 Denali (M = 7.9), 2003 San Simeon (M = 6.5), and 2004 Parkfield (M = 6.0) earthquakes. The observations consist of repeated GPS measurements of the position of one monument relative to another (separation ∼100 km). The early observations (t < 0.1 year) are well fit by the function a′ + c′log(t), where t is the time after the earthquake and a′ and c′ are constants chosen to fit the data. Because a log(t) time dependence is characteristic of transient (primary) creep, the early postseismic response may be governed by transient creep as Benioff proposed in 1951. That inference is provisional as the stress conditions prevailing in postseismic relaxation are not identical to the constant stress condition in creep experiments. The observed logarithmic time dependence includes no characteristic time that might aid in identifying the micromechanical cause.

Inverting for Slip on Three-Dimensional Fault Surfaces Using Angular Dislocations

The increasing quality of geodetic data (synthetic aperture radar interferometry [insar] dense Global Positioning System [gps] arrays) now available to geophysicists and geologists are not fully exploited in slip-inversion procedures. Most common methods of inversion use rectangular dislocation segments to model fault ruptures and therefore oversimplify fault geometries. These geometric simplifications can lead to inconsistencies when inverting for slip on earthquake faults, and they preclude a more complete understanding of the role of fault geometry in the earthquake process. We have developed a new three-dimensional slip-inversion method based on the analytical solution for an angular dislocation in a linear-elastic, homogeneous, isotropic, half-space. The approach uses the boundary element code Poly3D that employs a set of planar triangular elements of constant displacement discontinuity to model fault surfaces. The use of triangulated surfaces as discontinuities permits one to construct fault models that better approximate curved three-dimensional surfaces bounded by curved tiplines: shapes that commonly are imaged by three-dimensional reflection seismic data and inferred from relocated aftershock data. We demonstrate the method’s ability to model three-dimensional rupture geometries by inverting for slip associated with the 1999 Hector Mine earthquake. The resulting model avoids displacement anomalies associated with the overlapping rectangular dislocations used in previous models, improving the fit to the geodetic data by 32%, and honors the observed surface ruptures, thereby allowing more direct comparisons between geologic and geodetic data on slip distributions. Online Material : Hector Mine input files and file format description.

Nonlinear dynamics, granular media and dynamic earthquake triggering

The 1992 magnitude 7.3 Landers earthquake triggered an exceptional number of additional earthquakes within California and as far north as Yellowstone and Montana. Since this observation, other large earthquakes have been shown to induce dynamic triggering at remote distances--for example, after the 1999 magnitude 7.1 Hector Mine and the 2002 magnitude 7.9 Denali earthquakes--and in the near-field as aftershocks. The physical origin of dynamic triggering, however, remains one of the least understood aspects of earthquake nucleation. The dynamic strain amplitudes from a large earthquake are exceedingly small once the waves have propagated more than several fault radii. For example, a strain wave amplitude of 10(-6) and wavelength 1 m corresponds to a displacement amplitude of about 10(-7) m. Here we show that the dynamic, elastic-nonlinear behaviour of fault gouge perturbed by a seismic wave may trigger earthquakes, even with such small strains. We base our hypothesis on recent laboratory dynamic experiments conducted in granular media, a fault gouge surrogate. From these we infer that, if the fault is weak, seismic waves cause the fault core modulus to decrease abruptly and weaken further. If the fault is already near failure, this process could therefore induce fault slip.

Modal Property Changes of a Seismically Damaged Concrete Bridge

The Lavic Road Bridge in San Bernardino County, California, was severely damaged during the October 16, 1999, Hector Mine Earthquake (7.1 Mw). The earthquake epicenter was approximately 16 km (10 mi) south-southwest of the structure's location on Interstate 40. Prior to the earthquake, periodic modal field tests were being conducted on the structure to evaluate the impact of on-going reactive-aggregate induced structural deterioration. A scheduled field test was performed on the structure two weeks prior to the event. A second, unscheduled field test was performed three days after the event in order to capture any changes in the structure's modal properties that might have resulted from damage incurred as a result of the earthquake. This paper provides a visible description of the damage sustained by the structure, documents the field test procedures used to determine modal properties (pre-event and postevent modal frequencies, damping, and mode shapes), and discusses the changes observed in the pre-event and postevent modal properties. The extracted modal results indicated that lower-mode natural frequencies decreased 18% relative to pre-event values, that the frequency order of modes did not change, and that the modal damping levels increased by 25% relative to pre-event levels.

Response of desert pavement to seismic shaking, Hector Mine earthquake, California, 1999

The October 1999 Mw 7.1 Hector Mine earthquake in the Mojave Desert, California, generated characteristic surface disturbances on nearby desert pavements. These disturbances included (1) zones of wholesale gravel displacement interspersed with zones of intact pavement, (2) displaced and rotated cobbles, (3) moats around loosened, embedded boulders, (4) filling of abandoned cobble sockets, boulder moats, and other depressions with gravel, and (5) formation of narrow, subparallel, linear strips of exposed fine‐grained subpavement matrix (matrix lineations). Clasts displaced from matrix lineations and from cobble sockets tended to move downslope. Sharp boundaries of matrix lineations and slope‐controlled displacement directions on slopes of only a few degrees indicated that clasts remained close to the pavement surface during shaking. The regular, few decimeter spacing of matrix lineations suggests the presence of standing waves during seismic shaking. Boulder moats probably have good preservation potential and, at some desert pavement locations, might provide information on paleoseismic shaking. Although readily produced by coseismic shaking, displaced cobbles are unreliable indicators of past earthquake activity because of potential multiple origins. For an assumed earthquake recurrence interval of 10 ka, seismically driven sediment fluxes similar to those generated by the Hector Mine earthquake at the Lavic Siding pavement study site may be marginally competitive with aseismic smoothing mechanisms driven by bioturbation, rainbeat, and wash. For a 1 ka recurrence interval, seismic smoothing is likely to play a significant role in pavement evolution.

Onto what planes should Coulomb stress perturbations be resolved?

Coulomb stress maps are produced by computing the tensorial stress perturbation due to an earthquake rupture and resolving this tensor onto planes of a particular orientation. It is often assumed that aftershock fault planes are “optimally oriented”; in other words, the regional stress and coseismic stress change are used to compute the orientation of planes most likely to fail and the coseismic stress is resolved onto these orientations. This practice assumes that faults capable of sustaining aftershocks exist at all orientations, an assumption contradicted by the observation that aftershock focal mechanisms have strong preferred orientations consistent with mapped structural trends. Here we systematically investigate the best planes onto which stress should be resolved for the Landers, Hector Mine, Loma Prieta, and Northridge earthquakes by quantitatively comparing observed aftershock distributions with stress maps based on optimally oriented planes (two‐ and three‐dimensional), main shock orientation, and regional structural trend. We find that the best model differs between different tectonic regions but that in all cases, models that incorporate the regional stress field tend to produce stress maps that best fit the observed aftershock distributions, although not all such models do so equally well. Our results suggest that when the regional stress field is poorly defined, or in structurally complex areas, the best model may be to fix the strike of the planes upon which the stress is to be resolved to that of the main shock but allow the dip and rake to vary.

Evidence for non‐constant energy/moment scaling from coda‐derived source spectra

We present two lines of evidence showing the ratio of seismically radiated energy (ER) to seismic moment (Mo) increases with increasing moment for crustal events spanning 5 orders in seismic moment. This ratio, often referred to as the scaled energy (=ER/M0) has been elusive and the subject of recent debate within the seismological community. Because significant frequency‐dependent corrections must be made to account for source and path heterogeneity, it is not uncommon that energy estimates of the same event by different researchers have significantly different values. To minimize some of the potential problems, we used the regional coda methodology outlined by Mayeda et al. (2003) on 4 large earthquake sequences; Mw 7.4 and 7.2 Izmit/Duzce, Mw 7.2 Gulf of Aqaba, Mw 7.3 Landers, and Mw 7.2 Hector Mine. This methodology has been shown to provide the lowest variance estimate of the source spectrum when compared against traditional direct wave estimates. In our first approach we made energy estimates from coda‐derived spectra and observed that increases with increasing moment for 3.7 ≤ Mw ≤ 7.4. In our second approach we used the simple idea outlined by Prieto et al. (2004) to directly test for self similarity. Under self similarity, earthquake source spectra scaled by ω−3 should have the same shape. To test this, one needs very stable spectra either through spectral stacking of direct wave spectra or by the use of stable coda‐derived spectra. For all sequences we found that the mainshocks and larger aftershocks had significantly different spectral shapes than the smaller events. The scaled ω−3 spectra of the larger events represented an upper bound, having larger amplitudes near the corner frequency. In fact, an ω−3.5 scaling moved the larger events closer to the mean of the spectral population, though the spectral shape differences persisted. These results strongly suggest that earthquakes are not scaling self‐similarly and that complexities in the rupture process such as variable slip velocity exist between small and large events.