Converted Shear Waves Research Articles

Using 4C to characterize lithologies and fluids in clastic reservoirs

Multicomponent seismic has been around for a long time without gaining widespread acceptance. The potential benefits of having both compressional ( P -wave) and shear ( S -wave) information have never been disputed, but the cost of acquiring and processing the large data volumes has been high and the quality of the S -wave data has sometimes not met expectations. However, recent technological developments in the area of sources, sensors, telemetry, and data processing are making multicomponent data increasingly popular as a way to provide better quality data at lower cost. Although shear information can be accessed through the AVO response of the P -waves, the pure shear ( SS ) mode or the converted shear wave ( PS ) mode provide direct and more accurate information about the shear properties of the subsurface. Getting this improved information about the elastic properties of the subsurface is indeed one of the main reasons for acquiring multicomponent data. There are two major ways of acquiring multicomponent seismic data. Land multicomponent seismic is typically acquired using dedicated vertically and horizontally polarized traction sources to generate downgoing P - and S -wave modes. The seismic can then be recorded as three-component (3-C) or nine-component (9-C) data sets. The latter involves three consecutive 3-C recordings of direct and converted modes by activating each dedicated source mode in turn (i.e., vertical followed by in-line and cross-line horizontal components). Marine multicomponent data, known as 4-C, is acquired using the traditional air-gun sources to generate a pressure wave ( PP ), but the receivers are on the seafloor to record the 3-D vector field. The fourth component is a hydrophone recording the pressure field just like ordinary seismic streamers. By placing three-component geophones on the seafloor, converted shear waves, known as the PS mode, can also be recorded. A propagating wave front will, in addition to …

The scattering of low-grazing angle pulse beams at interfaces between fluid and porous media

Many environments in bottom interacting ocean acoustics consist of mud and sands which can be modeled as fluid saturated, porous solids with low shear moduli. To study the physical mechanisms responsible for forward- and backscattering in these media it is useful to have a forward modeling technique which applies to rough and heterogeneous porous bottoms. A numerical scattering chamber using the time-domain finite-difference method applied to the range-dependent Biot equations can be used to study the scattering of low-grazing angle pulse beams from fluid-saturated porous media. For example, for a flat seafloor over a homogeneous porous half-space, converted shear waves and converted compressional ‘‘slow’’ waves are shown to be excited in the subbottom even when the grazing angle is below the critical angle for compressional ‘‘fast’’ waves. Any scattering element near the seafloor will act as a secondary point source by Huygen’s principle and when excited by an acoustic wave will have the potential to generate a family of interface waves and body waves of ‘‘fast’’ compressional, ‘‘slow’’ compressional and shear type.

Impact of a complex overburden on analysis of bright reflections: A case study from the Mendocino Triple Junction

Determination of the physical properties and the geometry of features within the subsurface is a common application of active source seismic data. We use the example of unusually bright reflections identified in the lower crust of the Coast Ranges of northern California during the Mendocino Triple Junction Seismic Experiment to investigate the robustness of such determinations. The sources of these reflections are significant because they lie in a region of transition between subduction and strike‐slip tectonics, but they also have an overburden of heterogeneous Franciscan terrane rocks. We use finite difference synthetic seismograms including a stochastic overburden representative of Franciscan rocks and a realistic target geometry to examine how typical seismic diagonostics are affected under these conditions. The addition of a realistic overburden to the models replicates variations in the amplitude of the observed Pg arrival and leads to variations in measures of the amplitude of the reflected arrivals that are also similar to those in the data. Some of these variations are due to the measures of amplitude used, some represent real variations at the target, and others are due to distortion of the illumination field by the overburden. Overall reflection amplitudes are reduced due to transmission losses within the overburden, and the P wave seismic velocity within the reflectors may be as low as 2.5 km s−1 rather than the 3.5 km s−1 estimated when the overburden was not taken into account. Migrations of simulations, including combinations of overburden, target dip, and uneven spatial sampling, recover somewhat consistent but inaccurate approximations to the true target geometry: Discontinuous migrated images may result from a discontinuous wave field even where reflectors were continous. The simulations also imply that neither reversed reflection polarity nor converted shear waves are a reliable aspect of the reflected wave field if the overburden or target is complex. Comparison of migrated simulations to the data suggests that the reflections originate from two relatively continuous layers at the top and base of the lower crust, 300m and 100–150m in thickness, respectively. We believe that the reflections are from lenses of basaltic melt recently emplaced in the crust and generated by decompression melting in a “slab‐free” window.

Seismic Inversion and Deconvolution: Dual‐sensor Technology

Ocean bottom cables and dual‐sensor recordings at the sea floor are becoming increasingly popular tools for marine seismic exploration. They offer several advantages for seismic data processing, including multiple attenuation, ghost reflection removal, and direct measurement of converted shear waves at the sea floor. New techniques for processing these unconventional data are actively being developed.A book that specifically addresses the dual‐sensor technology written by Enders A. Robinson—one of the pioneers of seismic data processing—is therefore a welcome addition to the geophysical literature.

Using converted shear waves to image reservoirs with low-impedance contrast

Imaging sand/shale sequences is problematic when the impedance contrast between them is very small, a problem commonly encountered in basins with Tertiary clastics. Impedance trends for sand and shale in these basins display higher impedances for shales than for sands at shallow depths; however, sands exhibit higher impedances than shales at greater depths. That implies a crossover between the trends at some depth and that it is going to be difficult to image sands near the crossover depth with traditional seismic surveys. Sometimes stacking far-offset data is a solution because a strong AVO gradient can generate significant reflections at high angles of incidence, even if the normal-incidence reflection coefficient is close to zero. Converted waves can, in some cases, image the reservoir. The question is when because there is no generally accepted insight regarding what conditions favor converted-wave imaging. In other words, why does converted-wave imaging work very well in some reservoirs and not so well in others? Three conditions must exist for converted shear-wave imaging to be preferable to conventional P-P imaging: 1. A very small acoustic impedance contrast results in very weak reflections and thereby poor images. 2. The offset dependence of P-P amplitudes is not useful because (a) a small contrast in Poisson's ratio across the contact between seal and reservoir makes the P-P AVO gradient small, resulting in weak amplitudes at all offsets, or (b) velocity distribution in the overburden makes it impossible to achieve large angles of incidence at the reservoir level. 3. A large shear-wave impedance contrast generates strong converted waves. The first point simply states the obvious—converted-wave imaging is cost effective only when traditional P-P imaging fails, and P-P imaging fails when the acoustic-impedance contrast is close to zero. Point …

Early development of the southern Kerguelen Plateau (Indian Ocean) from shallow wide‐angle ocean bottom seismometer and multichannel seismic reflection data

We examine the early geological history of the southern Kerguelen Plateau (Indian Ocean) using ocean bottom seismometer (OBS), multichannel seismic (MCS), and Ocean Drilling Program data. Velocity‐depth models in the sedimentary Raggatt Basin are constrained by near‐range OBS data (refractions and reflections, including multiples and converted shear waves) and migrated MCS data. The models elucidate the significance of Lower Cretaceous lava flows, Albian to Coniacian/Santonian terrestrial and terrigenous sediment, and Maastrichtian and Paleocene seismic sequences. The Albian/Aptian basaltic basement complex consists of an upper and a lower series. The upper series is characterized by average Vps of 4.6–4.7 km/s, an upward increase of intercalated terrestrial sediment and altered flowtops, and a concomitant decrease in flow thickness; the lower series is marked by average Vps of 5.3–5.5 km/s, and contains thicker flows and less sediment. A volcanic center in the Raggatt Basin shows relatively low Vps (3.7–3.9 km/s for the upper series, 4.7–4.9 km/s for the lower series), and dipping reflections on the Raggatt Basin's flanks are also recorded as refractions. Terrestrial and terrigenous sediment of the Raggatt Basin, immediately overlying basement, is characterized by a seismic low‐velocity layer with Vps ranging from 2.2–2.9 km/s and a thickness of >1100 m in the central basin. Nearby source regions (e.g., Banzare Bank and other elevated areas south of the basin) account for a terrestrial and terrigenous sediment volume of ∼12,500 km3, deposited over ∼20 m.y. The depocenter of the Raggatt Basin began shifting in Santonian to early Maastrichtian time, and concluded by early late Paleocene time.

Poisson's ratio structure of young oceanic crust

We have applied full waveform inversion to wide‐aperture seismic reflection data from the southern East Pacific Rise near 14°S. The data contain clear compressional wave and doubly converted shear wave arrivals that provide good constraints on the P and S‐wave velocities (Vp, Vs, and hence Poisson's ratio σ) and seismic attenuation (Qp, Qs) structure of seismic layer 2. Layer 2A is highly attenuating (Qp = 18–30 and Qs = 8–15) and layer 2B is moderately attenuating (Qp = 30–50 and Qs = 20–25). Our results show high σ at the seafloor and in layer 2A (σ = 0.48). Across the top of the 2A/B transition the rapid increase in Vp is accompanied by a sharp drop in σ to 0.25 within just 200 m of the seafloor. We perform simple calculations to gain an insight into the porosity and crack distribution with depth. These calculations suggest that porosity is in excess of 30% in layer 2A but reduces to 6–7% at the top of the 2A/B transition and to about 5% at a depth of 600 m below seafloor within layer 2B. Our results suggest that there is an increase in the average aspect ratio with depth across the 2A/B transition. The most likely explanation is that numerous thin cracks either mechanically close or are infilled at depth. Our results show there to be an abrupt change in the pore structure across the 2A/B transition which is consistent with a lithologic transition from extrusives to dykes but is equally consistent with a transition (mechanical or hydrothermal) within the extrusive pile.

Estimation of reservoir fracturing from marine VSP using local shear‐wave conversion

A marine VSP is designed to estimate the orientation and density of fracturing within a gas‐producing dolomite layer in the southern North Sea. The overburden anisotropy is firstly estimated by analysing shear waves converted at or just below the sea‐bed, from airgun sources at four fixed offset azimuths. Full‐wave modelling helps confirm that the background has no more than 3% vertical birefringence, originating from TIH anisotropy with a symmetry axis orientated perpendicular to the maximum horizontal compressive stress of NW–SE. This finding concurs with current hypotheses regarding the background rock matrix in the upper crust. More detailed anisotropy estimates reveal two thin zones with possible polarization reversals and a stronger anisotropy. The seismic anisotropy of the dolomite is then determined from the behaviour of locally converted shear waves, providing a direct link with the physical properties of its fractures. It is possible to utilize this phenomenon due to the large seismic velocity contrast between the dolomite and the surrounding evaporites. Two walkaway VSPs at different azimuths, recorded on three‐component receivers placed inside the target zone, provide the appropriate acquisition design to monitor this behaviour. Anisotropy in the dolomite generates a transverse component energy which scales in proportion to the degree of anisotropy. The relative amplitudes, for this component, between the different walkaway azimuths relate principally to the orientation of the anisotropy. Full‐wave modelling confirms that a 50% vertical birefringence from TIH anisotropy with a similar orientation to the overburden is required to simulate the field observations. This amount of anisotropy is not entirely unexpected for a fine‐grained brittle dolomite with a potentially high fracture intensity, particularly if the fractures contain fluid which renders them compliant to the shear‐wave motion.

Thin Crust On the Flanks of the Slow-Spreading Southwest Indian Ridge

SUMMARY The Southwest Indian Ridge marks an end-member of mid-ocean ridge spreading rates, with a current full rate of only 13-15 mm a-'. Most thermal models of mid-ocean ridges suggest a decrease in crustal thickness at such slow spreading rates, because conductive heat loss from the upwelling asthenospheric mantle decreases the volume of melt generated by decompression. Seismic measurements of the thickness of crust formed at very slow-spreading ridges are sparse. We have reanalysed data from a twoship, split spread seismic refraction experiment conducted in 1962 on the southern flank of the Southwest Indian Ridge (Francis & Raitt 1967). We used synthetic seismograms to model amplitude variations, which were carefully recorded by the original investigators. 1- and 2-D modelling suggests that the seismic velocity increases smoothly with depth within the igneous crust, with an unusually high velocity of -6.0 km s--' near the top of the crust, increasing to -7.0 km s-' just above the Moho, and a crustal thickness of 5 km. Converted shear waves yield a Poisson's ratio of 0.30+0.01 in the crust, which is intermediate between values for gabbros and serpentinized upper mantle. The crustal thickness is consistent with a passive mantle upwelling model of melt generation at mid-ocean ridges. The unusual velocity structure may indicate the presence of gabbroic rocks near the seabed, unroofed by extension.

How does the shear-wave structure of the seabed affect the seismic wavefield?

SUMMARY Much modelling of the seismic wavefield is undertaken with an acoustic approximation in which the influence of shear waves is neglected. Although such calculations can predict the correct traveltimes for compressional-(P-) wave propagation, they can he very misleading with respect to the distribution of seismic amplitudes, especially at larger offsets. At the seabed, the acoustic approximation predicts total reflection for P waves incident beyond the critical angle. However, once the presence of shear waves in the solid material below the sea-floor is taken into account, P waves in the sea water incident beyond the critical angle can give rise to transmitted S waves, with a consequent major change in the propagation pattern. Such effects are very important for areas with high velocities at the sea-floor, as commonly occurs in tropical waters, such as the north-west shelf of Australia. The character of the water-borne noise in these conditions depends on whether the shear wavespeed at the seabed lies above or below the P-wave velocity in the sea water above. For high shear velocities, two distinct sets of critically reflected multiples can be produced to give a very energetic noise train trapped in the water column. Conversion of P to S in transmission at the sea-floor can often be important and give rise to significant arrivals on the outer traces from long marine cables. Further, the conversion of energy to S waves reduces the energy available for P-wave multiples and dramatically reduces the influence of waterbottom multiples compared with a purely acoustic situation. Synthetic seismogram calculations for large offsets and equivalent calculations in the slowness-times domain, with selective control of the level of multiples and conversion at each interface, provide a convenient tool for characterizing the expected water-borne energy and the influence of converted shear waves on the pressure field recorded in the water.

AVO modeling and the locally converted shear wave

The locally converted shear wave is often neglected in ray‐trace modeling when reproduction of the AVO response of potential hydrocarbon reservoirs is attempted. Primaries‐only ray‐trace modeling in which the Zoeppritz equations describe the reflection amplitudes is most common. The locally converted shear waves, however, often have a first‐order effect on the seismic response. This fact does not appear to be widely recognized, or else the implications are not well understood. Primaries‐only Zoeppritz modeling can be very misleading. Interference between the converted waves and the primary reflections from the base of the layers becomes increasingly important as layer thicknesses decrease. This interference often produces a seismogram that is very different from one produced under the primaries‐only Zoeppritz assumption. For primaries‐only modeling of thin layers, synthetic seismograms obtained by use of a linearized approximation to the Zoeppritz equations to describe the reflection coefficients are more accurate than those obtained by use of the exact Zoeppritz reflection coefficients. A real‐data example consisting of an assemblage of very thin layers has recently been discussed in the literature. Inferences as to the true earth properties based on the predicted amplitude variation with offset are in error because the primaries‐only assumption is invalid. For one of the models, primaries‐only modeling predicts an amplitude increase of approximately a factor of three from the near trace to the far trace. Reflectivity modeling predicts an amplitude decrease with offset. The O’Doherty‐Anstey effect suggests that transmission loss for primary reflections should not be included in normal‐incidence synthetic seismograms if the short‐period reverberations are not also included. The same principle holds for prestack modeling. Similarly, the Zoeppritz equations should not be used for synthetic seismograms without including the locally converted shear wave.

Poisson's ratio of a seaward-dipping reflector series, Hatton Bank

Seismic reflection profiles across the Hatton Bank continental margin in the eastern North Atlantic have imaged thick sequences of seaward-dipping reflectors. Converted shear waves propagating through one of these sequences were recorded on a two-ship expanding spread profile just landward of the ocean-continent transition. Synthetic seismogram modelling of these shear waves indicates that Poisson's ratio in the sequence is 0.27 ± 0.01, similar to that of normal oceanic Layer 2.

The Influence of the Shear-Wave Structure of the Sea Bed on the Seismic Wavefield

Much modelling of the seismic wave field is undertaken with an acoustic approximation in which the influence of shear waves is neglected. Although such calculations can predict the correct travel times for compressional (P) wave propagation, they can be very misleading with respect to the distribution of seismic amplitudes, especially at larger offsets. At the sea bed, the acoustic approximation predicts total reflection for P waves incident beyond the critical angle. However, once the presence of shear waves in the solid material of the sea floor is taken into account, P waves in the water can give rise to transmitted S waves with a major change in the propagation pattern. The effects are very important for hard sea bottoms such as those on the Northwest Shelf. The character of the water-borne noise in these conditions depends on whether the shear-wave speed at the sea bed lies above or below the P-wave velocity in the sea water above. For high shear velocities, two distinct sets of critically reflected multiples can be produced to give a very energetic noise train trapped in the water column. Conversion of P to S in transmission at the sea floor can often be important and give rise to significant arrivals on the outer traces from long marine cables. Synthetic seismogram calculations for large offsets, with selective control of the level of multiples and conversion at each interface, provide a convenient tool for characterizing the expected water-borne energy and the influence of converted shear waves on the pressure field recorded in the water.

Estimating SV‐wave stacking velocities for transversely isotropic solids

Conventional seismic experiments can record converted shear waves in anisotropic media, but the shear‐wave stacking velocities pose a problem when processing and interpreting the data. Methods used to find shear‐wave stacking velocities in isotropic media will not always provide good estimates in anisotropic media. Although isotropic methods often can be used to estimate shear‐wave stacking velocities in transversely isotropic media with vertical symmetry axes, the estimations fail for some transversely isotropic media even though the anisotropy is weak. For a given anisotropic medium, the shear‐wave stacking velocity can be estimated using isotropic methods if the isotropic Snell’s law approximates the anisotropic Snell’s law and if the shear wavefront is smooth enough near the vertical axis to be fit with an ellipse. Most of the 15 transversely isotropic media examined in this paper met these conditions for short reflection spreads and small ray angles. Any transversely isotropic medium will meet these conditions if the transverse isotropy is weak and caused by thin subhorizontal layering. For three of the media examined, the anisotropy was weak but the shear wave-fronts were not even approximately elliptical near the vertical axis. Thus, isotropic methods provided poor estimates of the shear‐wave stacking velocities. These results confirm that for any given transversely isotropic medium, it is possible to determine whether or not shear‐wave stacking velocities can be estimated using isotropic velocity analysis.

Processing, correlating, and interpreting converted shear waves from borehole data in southern Alberta

Borehole measurements coupled with phase information from Zoeppritz equation modeling has assisted in accurate correlation between a VSP converted S-wave section and both the surface and VSP P-wave sections from southern Alberta. For the most part, both the character and polarities of the sections agree; however, there are some differences. Some reflections are stronger and more distinct on the S-wave section than on the P-wave section. Spectral analysis of the time‐domain upgoing P-wave and S-wave energy shows that the frequency content of the S-waves is comparable to the P-waves. Thus, the slower velocity S-waves have a shorter wavelenght and provide better vertical resolution of some interfaces. Other upgoing S-wave modes can interfere with the P‐SV mode and contribute to the differences between the P- and S-wave sections. The match between P-wave and S-wave velocities ([Formula: see text] and [Formula: see text]), determined from VSP traveltime inversion and the full‐waveform sonic log, is best in the Paleozoic carbonate section; there is some discrepancy in Cretaceous sandstone intervals. A basal salt unit in the Paleozoic Beaverhill Lake formation has a VSP‐determined [Formula: see text] ratio of 1.97, suggesting that salt can be distinguished from carbonates using both P-wave and S-wave velocity information in this region.

Spectral Matrix Filtering Applied to Vsp Processing

The spectral matrix computed from VSP-traces transfer functions contains information about each wave making up the VSP data set. Using a filter based on the eigenvectors of the spectral matrix leads to a decomposition of input traces in eigensections. The eigensections associated with the largest eigenvalues contain the contribution of the correlated seismic events. Signal space is denoted as the sum of these eigensections. Other eigensections represent noise. When the different waves making up the VSP have very different amplitudes, decomposition of input traces into eigensections leads to wave separation without any required knowledge about the apparent velocities of the waves. Limitations of wave separation by the multichannel filtering are a function of the scalar product values of the waves (in frequency domain) and of the relative wave amplitudes. The spectral matrix filtering can always be used to enhance signal-to-noise ratio on VSP data. The eigenvalues of the spectral matrix can be used to estimate the signal-to-noise ratio as a function of frequency. It is possible to qualify the behavior of a VSP tool in a well and to detect some resonant frequencies probably generated by poor coupling. Field data examples are shown. The first example shows data recorded in a vertical well whose converted shear waves are separated from upgoing and downgoing compressional waves using a spectral matrix filter. This field case shows the efficiency of the spectral matrix filter in extracting weak events. The second example shows data recorded in a highly deviated well, where very close apparent velocity events are successfully separated by use of spectral matrix filtering.

MODELLING OF SOFT SEDIMENTS AND LIQUID‐SOLID INTERFACES: MODIFIED WAVENUMBER SUMMATION METHOD AND APPLICATION1

AbstractAlekseev and Mikhailenko have developed a wavenumber‐summation method which combines a finite integral transformation with a finite‐difference calculation and involves no approximations other than numerical ones. However, numerical anisotropy causes velocity errors for shear waves which are unacceptable if Poisson's ratios are larger than 0.4 and unless the number of grid points per wavelength is chosen considerably higher than the value generally regarded as sufficient in finite‐difference computations. To overcome this limitation in the applicability of the otherwise very powerful modelling scheme, the method is applied to the elastodynamic equations for the velocity vector. Thus, instead of solving a second‐order hyperbolic system as in the case of the wave equation, solutions to a first‐order hyperbolic system are computed. The finite‐difference iteration is performed in a staggered grid. In addition to mastering the numerical difficulties in cases where the Poisson's ratio is unusually high, this approach results in a code which can be used for the modelling of liquid layers.With the new scheme, water reverberations are investigated in terms of normal modes. It is found that for realistic sea‐bottom velocities the critical and supercritical cases exist only for P‐waves. It means that compressional waves are trapped within the water layer but energy leaks into the substratum through converted shear waves. These leaky compressional normal modes attain properties similar to those of shear normal modes or Pseudo‐Love waves. Due to their origin from conversion of dispersed multi‐modal compressional waves the shear waves generated at the sea‐bottom form a long complex wavetrain. They were found to mask the reflections from the target horizon in an offset‐VSP field section.

Acquisition and processing of pure and converted shear waves generated by compressional wave sources

Acquisition and processing of pure and converted shear waves generated by compressional wave sources

A Low-Velocity Zone Within the Layer 3 Region of 118 Myr Old Oceanic Crust In the Western North Atlantic

Summary During the North Atlantic transect (NAT) experiment 11 expanding spread profiles (ESPs) were acquired using a 3300 cubic inch, 30-element tuned airgun array (S/V Prospekta) and a 48-channel seismic streamer (R/V Moore). Extremely high-density seismograms were obtained that have allowed us to transform the observed data to the domain of intercept time and ray parameter (tau-p) with high precision. Velocity structures have been obtained by tau-sum inversion, ray tracing, and synthetic WKBJ and reflectivity seismogram methods both in tau-p and x-t domains. A detailed analysis of ESP 5 suggests the presence of a low-velocity zone within the lower oceanic crust at the location of magnetic anomaly MO (118 Ma). It is apparently 2.86km thick, has a p-wave velocity of 6.5 km s-1, S-wave velocity of 3.6 km s-1, and occupies much of the depth interval normally associated with oceanic layer 3. Converted shear wave arrivals have been used to obtain the shear velocity structure. We find that the Vp/Vs ratio in the uppermost part of the crust is about 1.9, and much of the crustal section has a Vp/Vsratio greater than √3. Attenuation of P- and S-waves has been studied by amplitude analysis, and we find that in most of the lower crust Qs is equal to Qp. Comparison with solutions from ESPs in younger and older parts of the basin suggests that ESP 5 is unique in displaying a low-velocity zone. Despite this the total thickness at ESP 5 is not significantly different than at other locations. We suggest that this provides strong evidence that the low-velocity zone occurs within the lower crust, and does not result from serpentinization of the upper mantle beneath thin oceanic crust.

Wave conversions from earthquakes and a new velocity model for the South Carolina coastal plain

Abstract Phase conversions from P to SV and from SV to P occur at a high impedance boundary near the surface in Charleston, South Carolina. Four arrivals (P, converted P, converted S, and S) are observed on three-component records of earthquakes in this area. Using arrival-time differences between paired arrivals of direct and converted phases, a shallow surface layer Vp/Vs ratio of 2.9 was determined. Applying the Wadati method to travel times derived at the base of the surface layer yields a Vp/Vs ratio in deeper layers of 1.73. Relocating earthquakes using this more appropriate velocity structure for direct and converted shear waves alters hypocentral parameters such that epicenters diverge and depths converge. It is inferred that these relocated earthquakes are not exclusively associated with a single seismogenic fault.