Similar Event Clusters Research Articles

Precise relative magnitude measurement improves fracture characterization during hydraulic fracturing

SUMMARY Microseismic monitoring is an important technique to obtain detailed knowledge of in-situ fracture size and orientation during stimulation to maximize fluid flow throughout the rock volume and optimize production. Furthermore, considering that the frequency of earthquake magnitudes empirically follows a power law (i.e. Gutenberg–Richter), the accuracy of microseismic event magnitude distributions is potentially crucial for seismic risk management. In this study, we analyse microseismicity observed during four hydraulic fracture treatments of the legacy Cotton Valley experiment in 1997 at the Carthage gas field of East Texas, where fractures were activated at the base of the sand-shale Upper Cotton Valley formation. We perform waveform cross-correlation to detect similar event clusters, measure relative amplitude from aligned waveform pairs with a principal component analysis, then measure precise relative magnitudes. The new magnitudes significantly reduce the deviations between magnitude differences and relative amplitudes of event pairs. This subsequently reduces the magnitude differences between clusters located at different depths. Reduction in magnitude differences between clusters suggests that some attenuation-related biases could be effectively mitigated with relative magnitude measurements. The maximum likelihood method is applied to understand the magnitude frequency distributions and quantify the seismogenic index of the clusters. Statistical analyses with new magnitudes suggest that fractures that are more favourably oriented for shear failure have lower b-value and higher seismogenic index, suggesting higher potential for relatively larger earthquakes, rather than fractures subparallel to maximum horizontal principal stress orientation.

Cliophysics: A scientific analysis of recurrent historical events

Named after Clio, the Greek goddess of history, cliophysics is a daughter (and in a sense an extension) of econophysics. Like econophysics it relies on the methodology of experimental physics. Its purpose is to conduct a scientific analysis of (recurrent) historical events. Such events can be of sociological, political or economic nature. In this last case cliophysics would coincide with econophysics. The main difference between cliophysics and econophysics is that the description of historical events may be qualitative as well as quantitative. For the handling of qualitative accounts cliophysics has developed an approach based on the identification of patterns. To detect a pattern the main challenge is to break the “noise barrier”. The very existence of patterns is what makes cliophysics possible and ensures its success. Briefly stated, once a pattern is detected, it allows predictions to be made. As the capacity to make successful predictions is the hallmark of any science, it becomes easy to decide whether or not the claim made in the title of the paper is indeed justified. A number of examples of clusters of similar events will be given which should convince readers that historical events can be simplified almost at will very much as in physics. In the last part of the paper, we describe cliophysical investigations conducted over the past decades; they make us confident that cliophysics can be a valuable tool for decision makers. This is not an ordinary physics paper for it does not focus on a specific question but rather delineates a new field and outlines an innovative research program. Neither should it be seen as an opinion paper for, thanks to its nine co-authors, it offers objective arguments in support of its proposals.

Along‐Dip Segmentation of the Slip Behavior and Rheology of the Copiapó Ridge Subducted in North‐Central Chile

AbstractWe studied the along‐dip influence of the Copiapó ridge subduction in the Atacama region, North‐Central Chile by building a new seismicity catalog, including similar events and non‐volcanic tremors (NVTs). We also obtained a 3‐D tomographic model for P and S‐waves velocity (and the implied Vp/Vs ratio). We identify along‐dip segmentation involving four distinct segments: a locked seismogenic zone hosting ordinary seismicity and clusters of similar events; a transition zone with NVTs and low seismicity; an aseismic zone with slow‐slip events; and a deep zone with abundant intraslab seismicity. The velocity models show differences among these zones, with low velocity anomalies of Vp and Vs coinciding with aseismic slip zones, indicating the possible presence of fluids. Due to the spatial distribution along‐strike and along‐dip of the aseismic zones, we propose that these differences in seismogenic behavior are generated by subduction of the heterogeneous seamounts associated with the Copiapó ridge.

GrowClust: A Hierarchical Clustering Algorithm for Relative Earthquake Relocation, with Application to the Spanish Springs and Sheldon, Nevada, Earthquake Sequences

Accurate earthquake locations are essential for providing reliable hazard assessments, understanding the physical mechanisms driving extended earthquake sequences, and interpreting fault structure. Techniques based on waveform cross correlation can significantly improve the precision of the relative locations of event pairs observed at a set of common stations. Here we describe GrowClust, an open‐source, relative relocation algorithm that can provide robust relocation results for earthquake sequences over a wide range of spatial and temporal scales. The method uses input differential travel times, cross‐correlation values, and reference starting locations, and applies a hybrid, hierarchical clustering algorithm to simultaneously group and relocate events within similar event clusters. The method is computationally efficient and numerically stable in its capacity to process large data sets and naturally applies greater weight to more similar event pairs. Additionally, it outputs location error estimates that can be used to help interpret the reliability and resolution of relocation results. As an example, we apply the GrowClust method to the recent Spanish Springs and Sheldon, Nevada, earthquake swarms. These sequences highlight the future potential for applying the GrowClust relocation method on a much larger scale within the region, where existing relocation results are sparse but vital for understanding the seismotectonics and seismic hazard of Nevada and eastern California.

Seismogenic magma intrusion before the 2010 eruption of Eyjafjallajökull volcano, Iceland

We present relatively relocated earthquake hypocentres for >1000 microearthquakes (ML < 3) that occurred during the 2 weeks immediately prior to the 2010 March 20 fissure eruption at Fimmvorðuhals on the flank of Eyjafjallajokull volcano in Iceland. Our hypocentre locations lie predominantly in horizontally separated clusters spread over an area of 10 km2 and approximately 4 km below sea level (5 km below the surface). Seismic activity in the final 4 d preceding the eruption extended to shallower levels <2 km below sea level and propagated to the surface at the Fimmvorðuhals eruption site on the day the eruption started. We demonstrate using synthetic data that the observed apparent ∼1 km vertical elongation of seismic clusters is predominantly an artefact caused by only small errors (0.01–0.02 s) in arrival time data. Where the signal-to-noise ratio was sufficiently good to make subsample arrival time picks by cross-correlation of both P- and S-wave arrivals, the mean depth of 103 events in an individual cluster were constrained to 3.84 ± 0.06 km. Epicentral locations are significantly less vulnerable to arrival time errors than are depths for the seismic monitoring network we used. Within clusters of typically 100 recorded earthquakes, most of the arrivals exhibit similar waveforms and identical patterns of P-wave first-motion polarities across the entire monitoring network. The clusters of similar events comprise repetitive sources in the same location with the same orientations of failure, probably on the same rupture plane. The epicentral clustering and similarity of source mechanisms suggest that much of the seismicity was generated at approximately static constrictions to magma flow in an inflating sill complex. These constrictions may act as a form of valve in the country rock, which ruptures when the melt pressure exceeds a critical level, then reseals after a pulse of melt has passed through. This would generate recurring similar source mechanisms on the same weak fault plane as the connection between segments of the sill system is repeatedly refractured in the same location. We infer that the magmatic intrusion causing most of the seismicity was likely to be a laterally inflating complex of sills at about 4 km depth, with seismogenic pinch-points occurring between aseismic compartments of the sills, or between adjacent magma lobes as they inflated. During the final 4 d preceding the eruption onset between 22:30 and 23:30 UTC on 2010 March 20, the seismicity suggests that melt progressed upwards to a depth of ∼2 km. This seismicity was probably caused by fracturing of the country rock at the margins of the propagating dyke. Subsequently, on the morning of the eruption a dyke propagated eastward from the region of precursory seismic activity to the Fimmvorðuhals eruption site.

Identification of faults activated during the stimulation of the Basel geothermal project from cluster analysis and focal mechanisms of the larger magnitude events

High-precision relative location procedures of the stronger seismic events (0.7≤ML≤3.4), based on cross-correlations of signals recorded by a six-sensor borehole network and numerous surface stations in the immediate epicentral area, show that clustering of hypocenters on different spatial scales is a dominant feature of the microseismicity induced by the stimulation of enhanced geothermal reservoir in Basel. In line with the fact that many of the observed earthquakes form clusters of similar events, several focal mechanisms are also nearly identical to each other. A comparison between the high-precision relative locations of the events within each cluster and the focal mechanisms often shows a good coincidence of the hypocentral distribution with one of the nodal planes of the focal mechanism. In some cases, the spatial extent of the individual clusters is limited to a few meters, which suggests that the corresponding events represent repeated slip with partial stress drop as pore pressures increase with time. In other cases, that include some of the stronger events (ML>2), the dimension of the individual clusters can amount to several 100m, and the activity within these clusters can extend over several days. Given that the orientation of many fault segments identified in this way deviates significantly from the overall orientation of the seismic cloud, these results reveal a complex internal structure of the flow paths in the rock volume stimulated by the water injection.

Indications for different types of brittle failure due to active coal mining using waveform similarities of induced seismic events

Abstract. Longwall mining activity in the Ruhr coal mining district leads to mining-induced seismicity. For detailed studies the seismicity of a single longwall panel beneath the town of Hamm-Herringen in the eastern Ruhr area was monitored between June 2006 and July 2007 with a dense temporary network of 15 seismic stations. More than 7000 seismic events with magnitudes between –1.7 ≤ ML ≤ 2.0 were detected and localized in this period. Most of the events occurred in the vicinity of the moving longwall face. In order to find possible differences in the brittle failure types of these events an association of the events to distinct clusters is performed based on their waveform characteristics. This task is carried out using a new clustering algorithm utilizing a network similarity matrix which is created by combining all available 3-component single station similarity matrices. The resultant network matrix is then sorted with respect to the similarity of its rows leading to a sorted matrix immediately indicating the clustering of the event catalogue. Finally, clusters of similar events are extracted by visual inspection. This approach results in the identification of several large clusters which are distinct with respect to their spatial and temporal characteristics as well as their frequency magnitude distributions. Comparable clusters are also found with a conventional single linkage approach, however, the new routine seems to be able to associate more events to specific clusters without merging the clusters. The nine largest observed clusters can be tentatively divided into three different groups that indicate different types of brittle failure. The first group consists of the two largest clusters which constitute more than half of all recorded events. Results of a relative relocation using cross-correlation data suggest that these events are confined to the extent of the mined out longwall and cluster close to the edges of the active longwall at the depth of active mining. These events occur in lockstep with the longwall advance and exhibit a high b value of the Gutenberg–Richter relation (GR) of about 1.5 to 2.5 and consist of small magnitude events. Thus, these events represent the immediate energy release adjacent to the mined out area. The second group consists of clusters located either slightly above or below the depth of active mining and occurring at the current position of the longwall face within the confines of the longwall. They consist of generally stronger events and do not follow GR. This activity might be linked to the failure of more competent layers above and below the mined out seam resulting in larger magnitude events. Finally, one cluster represents seismic activity with a rather low b value below 1 and events located partly towards the north of the longwall which are delayed with respect to the advance of the longwall face. These events are interpreted as brittle failure on pre-existing tectonic structures reactivated by the mining activity.

Seismic investigation of magmatic unrest beneath Mammoth Mountain, California, USA

I present a reinvestigation of the seismic activity during an 11-mo-long seismic swarm between 1989 and 1990 beneath Mammoth Mountain in eastern California (United States). This swarm, believed to be the result of a shallow intrusion of magma beneath Mammoth Mountain, was followed by the emission of magmatic CO 2 gas, resulting in tree die-off in 1990 and posing a signifi cant human health risk around Mammoth Mountain. In this study, I apply a state-of-the-art approach to estimate the ratio of compressional- to shear-wave velocity (Vp/Vs) within similar event clusters using differential times from waveform crosscorrelation. A high-resolution three-dimensional seismic velocity model and a precise earthquake location catalog are obtained to complement the analysis of the in situ Vp/Vs ratios. The majority of the relocated seismicity below sea level is distributed in an anomalous body with low Vp and high Vp/Vs, consistent with the seismic expression of the magma body that triggered the seismic swarm. Abrupt changes in Vp/Vs ratios, with lower values for the shallow depths, are observed near sea level, indicating the source of the 1989 CO 2 fl ux is a shallow CO 2 reservoir instead of the hypothesized basaltic magma at mid-crustal depths of 10-25 km. From sea level to the surface, low Vp/Vs ratios in a near-vertical zone suggest involvement of fl uid in the upward-migration of the seismicity.

Seismic velocity variations along the rupture zone of the 1989 Loma Prieta earthquake, California

We revisit the rupture zone of the 1989 Mw6.9 Loma Prieta earthquake, central California, by developing high‐resolution three‐dimensional (3‐D)Vp and Vp/Vs models. We apply the simul2000 inversion method and algorithm to a set of “composite” events, which have greater number of picks per event and reduced random picking errors compared with traditional master events. Our final P‐wave velocity model generally agrees with previous studies, showing a high velocity body of above 6.7 km/s in the southeast rupture zone of the main shock. The 3‐DVp/Vs model, however, has different features, with low Vp/Vs in the upper crust and high Vp/Vs anomalies in deeper layers of the rupture zone. We interpret the low Vp/Vs at shallow depths to be granitic rocks, whereas at greater depths the areas of higher Vp/Vs(around 1.725–1.75) presumably are mafic rocks. The resulting 3‐D velocity model was used to improve absolute locations for all local events between 1984 and 2010 in our study area. We then applied a similar event cluster analysis, waveform cross‐correlation, and differential time relocation methods to improve relative event location accuracy. Over 88% of the seismicity falls into similar event clusters. A dramatic sharpening of seismicity patterns is obtained after using these methods. The medians of the relative location uncertainties calculated by using the bootstrap approach are 5 m for horizontal and 8 m for vertical. Differential times from cross‐correlation are used to estimatein situnear‐sourceVp/Vsratio within each event cluster. The high‐resolutionVp/Vs method confirms the trend of the velocity variations from the tomographic results, although absolute values are slightly different.

Temporal Stability of Coda Q-1 in Southern California

Some studies of coda Q −1 have found temporal changes that may be associated with earthquake activity, but these analyses are subject to biases due to differences in source locations and other nonstationary behavior in earthquake catalogs. These biases can be greatly reduced by using clusters of repeating earthquakes; studies using this approach have generally found no resolvable changes in coda Q −1. We examine coda Q −1 variations across southern California using 22 similar event clusters identified from a recent large‐scale waveform cross‐correlation project to improve earthquake locations. These clusters are found across the region and span the time period between 1981 and 2005. We apply the method of Beroza et al. (1995) to compute differential coda Q −1 using waveforms from similar earthquake pairs and analyze the results to constrain any possible temporal variations. Results from individual event pairs show a great deal of scatter in differential coda Q −1, but exhibit no clear temporal variations or changes associated with large earthquakes. Application of a median filter to smooth the results shows that any persistent large‐scale changes in coda Q −1 during this time period are less than about 30%.

Non‐volcanic seismic swarms triggered by circulating fluids and pressure fluctuations above a solidified diorite intrusion

A non‐volcanic seismic swarm is often assumed to be tied to fluid movements, based on earthquake migrations obeying a diffusion equation. However, seismological observations that relate to the presence and driving force of migrating fluids are not well documented. Here we present high‐resolution tomographic imaging of seismic velocities in the Wakayama district, Japan, characterized by intense seismic swarms and high heat flow. Earthquake clusters at depths of 4–6 km are distributed within a zone with low‐Vp and anomalously low‐Vp/Vs. The existence of this low‐Vp/Vs anomaly is supported by another estimate using waveform cross‐correlation data from similar event clusters. We suggest that these anomalies are caused by fluid‐filled pores with relatively high aspect ratios. Just beneath the clusters of earthquakes, a prominent high‐velocity zone is clearly imaged. It lacks earthquakes, and could represent a body of intruded diorite related to past volcanism. We propose that non‐volcanic seismic swarms are triggered by circulating fluids and fluid pressure fluctuations driven by the thermal anomaly of the solidified diorite magma.

Evidence for water‐filled cracks in earthquake source regions

We identify lowered Vp/Vs ratios near earthquake source regions in southern California using observations from a seismic tomography model and high‐resolution local Vp/Vs estimates using waveform cross‐correlation data from within similar event clusters. The median tomographic Vp/Vs ratio is 1.716 ± 0.008 at all the relocated crustal earthquake locations, compared to the background median value of 1.729 ± 0.007 for the tomography model, although the error estimates overlap slightly. The median in situ Vp/Vs ratio of 1.673 ± 0.022 within the similar event clusters suggests that tomographic studies are overestimating Vp/Vs at source regions. Interpretation of Vp/Vs anomalies is complicated by the scatter in values obtained for individual clusters and in comparisons to absolute Vp and Vs velocities in the tomography model. However, the low Vp/Vs ratios measured for the seismicity clusters are hard to explain with known rocks and suggest the presence of water‐filled cracks with several percent porosity in earthquake source regions in southern California, which likely has an effect on faulting and earthquake activity.

A Search for Temporal Variations in Station Terms in Southern California from 1984 to 2002

We use relative arrival times and locations for similar earthquake pairs that are found using a cross-correlation method to analyze the time dependence of P and S station terms in southern California from 1984 to 2002. We examine 494 similar event clusters recorded by Southern California Seismic Network (SCSN) stations and compute absolute arrival-time variations from the differential arrival-time resi- duals obtained following event relocation. We compute station terms from the robust means of the absolute arrival-time residuals from all events recorded by each station at 3-month intervals. We observe nine stations with abrupt offsets in timing of 20- 70 msec, which are likely caused by equipment changes during our study period. Taking these changes into account could improve the relative location accuracy for some of the event clusters. For other stations, we generally do not see systematic temporal variations greater than about 10 msec. Analysis of residuals along individual ray paths does not reveal any clear localized regions of apparent velocity changes at depth. These results limit large-scale, long-lasting temporal variations in P and S ve- locities across southern California during this time period to less than about � 0:2%. However, there is an increased fraction of individual travel-time residuals exceeding 20 msec immediately following major earthquakes from source regions near the main- shock rupture.

Applying a three‐dimensional velocity model, waveform cross correlation, and cluster analysis to locate southern California seismicity from 1981 to 2005

We compute high‐precision earthquake locations using southern California pick and waveform data from 1981 to 2005. Our latest results are significantly improved compared to our previous catalog by the following: (1) We locate events with respect to a new crustal P and S velocity model using three‐dimensional ray tracing, (2) we examine six more years of waveform data and compute cross‐correlation results for many more pairs than our last analysis, and (3) we compute locations within similar event clusters using a new method that applies a robust fitting method to obtain the best locations satisfying all the differential time constraints from the waveform cross correlation. These results build on the relocated catalogs of Hauksson and Shearer (2005) and Shearer et al. (2005) and provide additional insight regarding the fine‐scale fault structure in southern California and the relationship between the San Andreas Fault (SAF) and nearby seismicity. In particular, we present results for two regions in which the seismicity near the southern SAF seems to align on dipping faults.

Estimating Local Vp/Vs Ratios within Similar Earthquake Clusters

We develop and test a method to estimate local Vp / Vs ratios for compact similar earthquake clusters using the precise P and S differential times obtained using waveform cross-correlation. We demonstrate how our technique works using synthetic data and evaluate likely errors arising from near-source takeoff angle differences between P and S waves. We use a robust misfit function method to compute Vp / Vs ratios for both synthetic data sets and several similar event clusters in southern California, and use a bootstrap resampling approach to estimate standard errors for real data. Our technique has higher resolution for near-source Vp / Vs ratios than typical tomographic inversion methods and provides constraints on near-fault rock properties.

Self-similarity patterns of precipitation in the Iberian Peninsula

Summary An objective classification of the precipitation field over the Iberian Peninsula and the Balearic Islands is obtained. Data are derived from a high-resolution daily precipitation dataset obtained from in-situ measurements. The dataset, Iberian monthly Precipitation Dataset (IPD), consists of monthly precipitation data over a 25 km � 25 km grid from 1 st January 1961 to 31 st December 2003. Therefore, 960 data series over the Iberian Peninsula and the Balearic Islands are disposed over the grid for 43-year period. Multi-resolution wavelet analysis is used to extract similar information in the precipitation field at different timescales. An objective classification of the obtained wavelet coefficient series is carried out by means of the Kohonen’s neural network, also named Self-Organizing Map (SOM). SOM is formed by an unsupervised learning algorithm that may be used to find clusters of similar events in the input data and is able to identify some underlying dynamic structures of the multi-dimensional datasets. SOM is applied to the wavelet coefficients for intramonthly, intermonthly and interannual oscillations, obtaining self-organised maps which objectively identify similar zones of precipitation behaviour over the Iberian Peninsula. The homogeneity of the patterns is also studied by means of non-parametric correlations, energy scalograms and tests of significance. The intramonthly, intermonthly and interannual waves resulted in seven, five and three SOM patterns, respectively. As timescale increases, the wavelet series coefficients tend to be highly clustered. The results indicate that as the oscillation frequencies decrease, the Iberian precipitation behaves more linearly.

Southern California Hypocenter Relocation with Waveform Cross-Correlation, Part 2: Results Using Source-Specific Station Terms and Cluster Analysis

We obtain precise relative relocations for more than 340,000 southern California earthquakes between 1984 and 2002 by applying the source-specific station-term (SSST) method to existing P- and S-phase picks and a differential lo- cation method to about 208,000 events within similar-event clusters identified with waveform cross-correlation. The entire catalog is first relocated by using existing phase picks, a reference 1D velocity model, and the SSST method of Richards-Dinger and Shearer (2000). We also perform separate relocations of Imperial Valley events by using a velocity model more suited to this region. Next, we apply cluster analysis to the waveform cross-correlation output to identify similar-event clusters. We re- locate earthquakes within each similar-event cluster by using the differential times alone, keeping the cluster centroid fixed to its initial SSST location. We estimate standard errors for the relative locations from the internal consistency of differential locations between individual event pairs; these errors are often as small as tens of meters. In many cases the relocated events within each similar-event cluster align in planar features suggestive of faults. We observe a surprising number of such faults at small scales that strike nearly perpendicular to the main seismicity trends. In general, the fine-scale details of the seismicity reveal a great deal of structural com- plexity in southern California fault systems.

Spatiotemporal variations of crustal anisotropy from similar events in aftershocks of the 1999M7.4 İzmit andM7.1 Düzce, Turkey, earthquake sequences

We analyse spatiotemporal variations of crustal anisotropy along the Karadere–Düzce branch of the North Anatolian Fault from similar earthquakes in the aftershock regions of the 1999 Mw7.4 İzmit and Mw7.1 Düzce earthquakes. The similar earthquake clusters are identified by performing cross-correlation on waveforms generated by ∼18 000 earthquakes. Depending on the applied similarity criterion, about 4–60 per cent of the events belong to similar event clusters. Splitting parameters averaged within each cluster show significant variations for slightly different ray paths, indicating strong spatial variations of crustal anisotropy in this area. We also find clear changes in the spatiotemporal seismicity patterns following the Düzce main shock. Large apparent co-seismic changes (up to 30 per cent) of shear wave splitting delay times are observed across the time of the Düzce main shock at stations near the epicentral region. However, the changes can be mostly explained by the spatial variations of ray paths due to the changing seismicity, rather than changes in the properties of the anisotropic medium. Splitting parameters measured within similar earthquake clusters indicate at most 2 per cent changes in delay times associated with the occurrence of the Düzce main shock. The results do not show systematic precursory changes before the Düzce main shock.

Characteristics of deep (≥13 km) Hawaiian earthquakes and Hawaiian earthquakes west of 155.55°W

High precision relocation of earthquakes recorded by the Hawaiian Volcano Observatory (HVO) seismic network provides new information on the characteristics of seismic faulting at this oceanic hot spot. Using waveform cross correlation, we have measured correlation coefficients and travel time differences for a set of 14,605 deep (≥13 km) earthquakes recorded from 1988 to 1998. We find that about half of the analyzed earthquakes are in similar event clusters that delineate fault zones in the lower crust and upper mantle. We suggest that much of this deep seismicity reflects rupture in the brittle lithosphere away from the magma pathways, although at Kilauea the stresses from magma movement may additionally help trigger mantle earthquakes on preexisting faults in regions with high differential ambient stresses. Focal mechanisms of similar event clusters throughout Hawaii display characteristic patterns and appear consistent with the hypothesis that deep earthquakes on preexisting faults reflect the stresses due to volcano loading and flexure. We also present the results of applying cross correlation analyses and relocation to ∼7000 earthquakes at all depths located west of 155.55°W and recorded from 1988 to 1998. The pattern of relocated earthquakes at the Kealakekua fault zone is consistent with the presence of a low‐angle detachment on the west flank of Mauna Loa.

Analysis of similar event clusters in aftershocks of the 1994 Northridge, California, earthquake

We perform waveform cross‐correlation on over 14,000 aftershocks of the 1994 Northridge MW 6.7 earthquake in southern California as recorded by short‐period stations of the Southern California Seismic Network (SCSN). Approximately 10–30% of the events belong to similar event clusters, depending upon the similarity criteria that are applied. We relocate events within 218 of these clusters to a relative location accuracy of about 30 m using the differential times obtained from the cross‐correlation. These relocated event clusters often show planar features suggestive of faults at depth, and we apply principal parameter analysis to characterize the shape of each cluster and to compute best fitting planes. In several cases, these planes are parallel to the main shock fault plane; however, more generally, the seismicity planes exhibit a wide range of orientations, suggesting complexity in the aftershock faulting. Composite focal mechanisms can be obtained for each cluster by combining the P polarity data from individual events. A comparison of polarity measurement differences within similar event clusters provides constraints on the error rate in the individual focal mechanisms. For some clusters, we are able to resolve the primary versus auxiliary fault plane ambiguity by comparing the computed focal mechanisms with the best fitting seismicity planes. Individual event focal mechanisms are in general agreement with the composite focal mechanisms for the similar event clusters. Events occurring along the main shock rupture plane are mainly thrust, whereas events in the hanging wall are predominately strike‐slip.