Search

We investigate possible links between seismicity and fluid injection in the Williston basin in the north‐central United States, focusing on the region around the Bakken formation unconventional hydrocarbon play. Here, we show earthquakes are rarer near injection wells in the Williston basin than in the Fort Worth basin of Texas or in central Oklahoma. To identify earthquakes, we analyze seismograms collected by EarthScope USArray temporary stations, deployed on a grid with 70 km spacing. During the September 2008–May 2011 study period, we identified only nine regional earthquakes; of these only three were situated near injection wells. The reason why Williston basin earthquakes are so scarce is unclear. In both the Bakken and Barnett Shale play regions, injection volumes increased significantly in late 2007, and both areas have very low levels of natural seismicity. Oklahoma has experienced much higher rates of apparently induced seismicity than either region, possibly because injection volumes are higher in some wells in Oklahoma.

Between November 2009 and September 2011 the EarthScope USArray program deployed ∼25 temporary seismograph stations on a 70-km grid in south-central Texas between 27°N–31°N and 96°W–101°W. This area includes the Eagle Ford Shale. For decades this geographic region has produced gas and oil from other strata using conventional methods, but recent developments in hydrofracturing technology has allowed extensive development of natural gas resources from within the Eagle Ford. Our study surveys small-magnitude seismic events and evaluates their correlation with fluid extraction and injection in the Eagle Ford, identifying and locating 62 probable earthquakes, including 58 not reported by the U.S. Geological Survey. The 62 probable earthquakes occur singly or in clusters at 14 foci; of these foci, two were situated near wells injecting recently increased volumes of water; eight were situated near wells extracting recently increased volumes of oil and/or water; and four were not situated near wells reporting significant injection/extraction increases. Thus in this region, while the majority of small earthquakes may be triggered/induced by human activity, they are more often associated with fluid extraction than with injection. We also investigated the M W 4.8 20 October 2011 Fashing earthquake—the largest historically reported earthquake in south-central Texas—that occurred two weeks after the removal of the temporary USArray stations. A field study indicated that the highest-intensity (MMI VI) region was about 10 km south of 2010–2011 foreshock activity, and that there were no high-volume injection wells within 20 km of the MMI V–VI region or the foreshocks. However, the 20 October 2011 earthquake did coincide with a significant increase in oil/water extraction volumes at wells within the MMI V–VI region, and this was also true for previous earthquakes felt at Fashing in 1973 and 1983. In contrast, our study found significant increases in injection prior to an m bLG 3.6 20 July 1991 earthquake near Falls City, Texas. Thus the Eagle Ford geographic region, with seismic activity associated both with extraction and injection, appears to be more complex than the Barnett Shale of northeast Texas, where a similar survey found possible correlations only with fluid injection.

On 9 June 2009, an ![Graphic][1] 2.8 earthquake shook Cleburne, Texas, a community not known to have previously experienced earthquakes. Over 50 small earthquakes followed by the end of December 2009. A temporary network of four and then five IRIS‐Passcal broadband systems was deployed from June 2009 to June 2010, recording data that were used to locate 38 events with the most confident P ‐ and S ‐arrival picks. Event locations were distributed along a 2 km long north‐northeast trend. The location centroid was at 32.298° N, 97.372° W and at 3.6 km depth. This location is approximately 1.3 km from a saltwater disposal well that began injection in October 2007 and 3.2 km away from a second injection well that was active from September 2005 to late July 2009. Focal mechanisms estimated for the best‐recorded events suggest a north‐northeast–south‐southwest‐trending normal fault with a dip of ∼50° and a component of oblique motion (rake of ∼−80°). This average solution is generally consistent with the north‐northeast‐trending extensional faults that are prevalent across parts of Texas, Oklahoma, Louisiana, and Arkansas. Stress drops calculated from P and S spectra for seven different events ranged from 3.9 to 90 bars, with most estimates between 40 and 50 bars, typical values for intraplate earthquakes. Because there were no known previous earthquakes, and the located events were close to the two injection wells and near the injection depth, the possibility exists that earthquakes may be related to fluid injection. Online Material: Tables of absolute and relative earthquake locations and a figure summarizing focal mechanism solutions. [1]: /embed/inline-graphic-1.gif

Between November 2009 and September 2011 the EarthScope USArray program deployed ∼25 temporary seismograph stations on a 70-km grid in south-central Texas between 27°N–31°N and 96°W–101°W. This area includes the Eagle Ford Shale. For decades this geographic region has produced gas and oil from other strata using conventional methods, but recent developments in hydrofracturing technology has allowed extensive development of natural gas resources from within the Eagle Ford. Our study surveys small-magnitude seismic events and evaluates their correlation with fluid extraction and injection in the Eagle Ford, identifying and locating 62 probable earthquakes, including 58 not reported by the U.S. Geological Survey. The 62 probable earthquakes occur singly or in clusters at 14 foci; of these foci, two were situated near wells injecting recently increased volumes of water; eight were situated near wells extracting recently increased volumes of oil and/or water; and four were not situated near wells reporting significant injection/extraction increases. Thus in this region, while the majority of small earthquakes may be triggered/induced by human activity, they are more often associated with fluid extraction than with injection. We also investigated the M W 4.8 20 October 2011 Fashing earthquake—the largest historically reported earthquake in south-central Texas—that occurred two weeks after the removal of the temporary USArray stations. A field study indicated that the highest-intensity (MMI VI) region was about 10 km south of 2010–2011 foreshock activity, and that there were no high-volume injection wells within 20 km of the MMI V–VI region or the foreshocks. However, the 20 October 2011 earthquake did coincide with a significant increase in oil/water extraction volumes at wells within the MMI V–VI region, and this was also true for previous earthquakes felt at Fashing in 1973 and 1983. In contrast, our study found significant increases in injection prior to an mbLG3.6 20 July 1991 earthquake near Falls City, Texas. Thus the Eagle Ford geographic region, with seismic activity associated both with extraction and injection, appears to be more complex than the Barnett Shale of northeast Texas, where a similar survey found possible correlations only with fluid injection.

This paper evaluates neotectonic segmentation in the Solomon Islands forearc, and considers how it relates to regional tectonic evolution and the extent of ruptures of large megathrust earthquakes. We first consider regional geomorphology and Quaternary vertical displacements, especially uplifted coral reef terraces. Then we consider geographic seismicity patterns, aftershock areas and vertical displacements for large earthquakes, focal mechanisms, and along-arc variations in seismic moment release to evaluate the relationship between neotectonically defined segments and seismicity. Notably, one major limitation of using seismicity to evaluate arc segmentation is the matter of accurately defining earthquake rupture zones. For example, shoreline uplifts associated with the 1 April 2007 Mw 8.1 Western Solomons earthquake indicate that the along-arc extent of rupture was about 50 km smaller than the aftershock area. Thus if we had relied on aftershocks alone to identify the 2007 rupture zone, as we do for most historical earthquakes, we would have missed the rupture's relationship to a major morphologic feature. In many cases, the imprecision of defining rupture zones without surface deformation data may be largely responsible for the poor mismatches to neotectonic boundaries. However, when a precise paleoseismic vertical deformation history is absent, aftershocks are often the best available tool for inferring rupture geometries. Altogether we identify 16 segments in the Solomon Islands. These comprise three major tectonic regimes or supersegments that correspond respectively to the forearc areas of Guadalcanal–Makira, the New Georgia island group, and Bougainville Islands. Subduction of the young and relatively shallow and buoyant Woodlark Basin and spreading system distinguishes the central New Georgia supersegment from the two neighboring supersegments. The physiographic expression of the San Cristobal trench is largely absent, but bathymetric mapping of the surface trace of the interplate thrust zone defines it adequately. The New Georgia supersegment has smaller arc segments, and more islands due to general late Quaternary forearc uplift very close to the trench where vertical displacement rates tend to be faster; prior to the 2007 earthquake it had much lower rates of seismic activity than the neighboring supersegments. Generally the mean along-arc lateral extent of Solomon arc segments is about 75 km, somewhat smaller than the segments reported in some other island arcs such as Japan (~ 100–260 km), but larger than those of the Tonga (30–80 km) and Central New Hebrides arcs (30–110 km). These differences may be real but it may occur simply because the coral-friendly tropical environment of the South Pacific arcs, numerous emerged forearc islands, and high seismicity rates provide an unusually favorable situation for observing variations in vertical tectonic activity and thus for identifying segment boundaries. Over the past century seismic slip in the Solomons, as indicated by seismic moment release, has corresponded to about half the plate convergence rate; however, there are notable variations along the arc. Even with the 2007 earthquake, the long-term moment release rate in the New Georgia supersegment is relatively low, and this may indicate that large earthquakes are imminent.

Temperature profoundly affects seismicity and rheology within subducted lithosphere; the thermal parameter Φ is often used to characterize the temperature structure of subduction zones. Φ is usually defined as tAgeV⊥, where tAge is the age of the plate as subduction begins and V⊥ is the lithosphere's vertical descent rate. But, interpretations of Φ can be confusing because its relationship with temperature is seldom expressed explicitly and because published estimates of Φ often vary by factors of two or more. This paper presents a particularly simple approximate model that leads to an analytical relationship between Φ/h and the minimum temperature within subducted lithosphere at a depth h. In spite of the model's simplicity, it gives reasonable approximate temperature estimates, e.g., for subduction zones where Φ is less than about 2500 km, the relationship between Φ and the observed maximum depth hmax of seismicty is consistent with the observation that seismicity ceases when interior slab temperature exceeds about 600–800 °C.

There are at least 30 major plate boundary segments worldwide where the plate boundary changes from subduction to strike-slip; these include six triple junctions and 24 two-plate boundaries. This study investigates earthquake seismicity in the 24 two-plate subduction-to-strike-slip transition (SSST) regions by utilizing recently published earthquake relocations, ternary diagrams of focal mechanisms, and moment rate calculations. To facilitate cross-regional comparisons, we categorize the geometry of SSST plate boundaries in terms of (1) their radius of curvature, (2) their sense of curvature, that is, whether they are convex or concave as viewed from the downgoing plate, and (3) their tectonic complexity, that is, the variability of crustal thickness and the segmentation of the plate boundary trace. We observe three main trends in SSST regions: (1) there is a conspicuous scarcity of strike-slip earthquakes along plate boundary segments that plate motion models indicate are strike-slip boundaries; (2) in these apparent strike-slip segments, both the rate of occurrence of earthquakes of any kind and the moment release rate are low compared to adjacent subduction segments; and (3) there were few observable differences in seismicity between convex and concave boundaries. The observation that transform zones exhibit moment rate deficiencies, that is, have few large-magnitude earthquakes in the historical record, may have important implications for seismic hazard assessment in SSST regions. In particular, is motion along these boundaries aseismic with little seismic hazard, or is motion expressed in very large magnitude, infrequent, but potentially devastating earthquakes? In at least three such regions, New Zealand, the Philippines, and the Dominican Republic, paleoseismic evidence and the historical record of seismicity suggest that very large, infrequent earthquakes do occur.

Summary This paper describes quantitative methods for evaluating the distribution of focal mechanism orientations of groups of earthquakes, and shows how to display this distribution on a triangle diagram. We present a χ2-based statistical test for determining whether two sets of focal mechanisms are drawn from distinct populations. We apply these methods to 3625 better-determined mechanisms for shallow earthquakes in the Harvard Centroid Moment Tensor (CMT) catalogue; we describe the distributions of mechanisms in the whole catalogue and for catalogue subsets in some specific tectonic environments. In addition, we explore the geographical locations of mechanisms with orientations that occur relatively infrequently, that is, mechanisms that are unlike thrust, normal, or strike-slip mechanisms. Such mechanisms are relatively rare along mid-ocean ridges and in oceanic subduction zones. The majority of these unusual mechanisms occur along plate boundaries where crustal thickness is highly variable, and in regions where the plate convergence direction becomes oblique and thus relative motion changes from convergence to transform motion.

New marine geophysical data along the Macquarie Ridge Complex, the Australia-Pacific plate boundary south of New Zealand, illuminate regional neotectonics. We identify tectonic spreading fabric and fracture zones and precisely locate the Australia-Pacific plate boundary along the Macquarie Ridge Complex. We interpret a ∼5–10 km wide Macquarie Fault Zone between the two plates along a bathymetrie high that extends nearly the entire length of the Australia-Pacific plate boundary south of New Zealand. We conclude that this is the active Australia-Pacific strike-slip plate boundary. Arcuate fracture zones become asymptotic as they approach the plate boundary. A broad zone of less intense deformation associated with the plate boundary extends ∼50 km on either side of the Macquarie Fault Zone. Marine geophysical data suggest that distinct segments of the plate boundary have experienced convergence and strike-slip deformation, although teleseismic evidence overwhelmingly indicates strike-slip motion along the entire surveyed boundary today. The McDougall and southernmost Puysegur segments show no evidence for past underthrusting, whereas data from the Macquarie and Hjort segments strongly suggest past convergence. The present-day strike-slip plate boundary along the Macquarie Ridge Complex coincides with the relict spreading center responsible for Australia-Pacific crust in the region. Our conceptual model for the transition from seafloor spreading to strike-slip motion along the Macquarie Ridge Complex addresses the decreasing length of spreading center segments and spacing between fracture zones, as well as the arcuate bend of the fracture zones that become asymptotic to the current transform plate boundary.

Traveltime-based location methods are ineffective for locating regional seismic events which are recorded by few stations; however, the Comprehensive Test Ban Treaty (CTBT) makes it imperative that we monitor sparsely observed events. This paper reviews the current status of single-station location (SSL) methods—the location of seismic events using records from one three-component broadband digital station. We argue here that SSL methods may ultimately provide the most effective and economical means for monitoring small-magnitude (M∼3) seismic events over most of the earth. Potentially, station-event distance and focal depth may be determined accurately by utilizing broadband three-component data and matching observed and synthetic waveforms; station-event azimuth is determinable from the polarization of the P-wave arrival and P coda. However, more research is required before SSL methods become a practical alternative to traveltime-based methods. Currently, the most significant problem is that regional seismograms are strongly dependent on regional crustal structure. Thus, future research efforts should focus both on determining appropriate crustal models in specific regions, and also on improved methods for dealing with regional variations in crustal structure. A second important problem concerns location errors; with SSL they may involve 180° errors in station-event azimuth or other highly non-Gaussian `blunders'.

SUMMARY In this paper we describe a practical method for determining crustal structure by comparing observations of teleseismic body waves with synthetics constructed using generalized ray theory. Layered crustal structure beneath a seismic station strongly influences teleseismic body waves arriving in the 50 s or so after the initial P; however, the effect of crustal structure differs significantly for vertical-component and radialcomponent records. For any proposed crustal structure, we develop an approximate theoretical formulation to predict the radial-component seismogram as a function of the vertical-component observations. Then, we compare the observed and predicted radial-component seismograms in the time domain, and vary the proposed crustal structure using a grid-search scheme until we obtain an optimum match. The theoretical approximation is complete to second order in the reflection and transmission coefficients, and is accurate to within a few per cent for modelling teleseismic waveforms. We demonstrate the feasibility of this second-order, radial-vertical comparison (SORVEC) method by testing it on synthetically generated waveforms having complicated source properties and significant levels of background noise. We illustrate the application to real data by determining the crustal structure beneath two seismic stations in Tibet, LZH and AMDO. Beneath station LZH, the crustal thickness is about 65 km. Beneath station AMDO, the crustal thickness is about 75 km, and there is a low-velocity layer in the lower crust.

INTRODUCTION As earthquake seismologists, we routinely plot “beachball” focal mechanisms describing the pattern of P -wave first-motions for ordinary, garden-variety double-couple earthquakes. But, how does one plot the pattern for Sh and Sv waves or for earthquake sources with isotropic or compensated linear vector dipole (CLVD) components? This is one of those questions that has always threatened me with a feeling of vague discomfort because I was sure I should know the answer, but I wasn't sure I did, even though I thought it was probably easy. Recently I needed to know the answer for two reasons: to check some software I had written for constructing synthetic seismograms and to determine whether observations of S waves at certain specific seismic stations had the same sign as the radiation predicted by the published Harvard centroid moment tensors (CMT). So, I resolved to spend a “few hours” writing a program to plot nodal...

In this paper we investigate the value of b in the Gutenberg-Richter relation for four teleseismic catalogs of earthquakes: Abe's historical catalog, the Harvard Centroid Moment Tensor (CMT) catalog, the catalog of the International Seismological Centre (ISC), and the Blacknest catalog. An unfortunate result is that b differs by 30% or more when determined in different magnitude ranges, in different catalogs, or using different methods. For global catalogs separated into shallow, intermediate, and deep earthquake groups, all values determined for b lie between 0.72 and 1.34. We can identify no systematic global variation of b with depth. For teleseismic catalogs it is difficult to believe measured geographic variations in b because systematic errors cause problems of earthquake detection, earthquake location, aftershock identification, and magnitude determination. However, some variations in b are so persistent and large that they must be real. For deep earthquakes in Tonga-Fiji, for example, various measurements of b He between 1.06 and 1.57, comparable to b for shallow earthquakes, whereas measurements of b for deep earthquakes in the rest of the world are much lower, between 0.53 and 0.96. For shallow earthquakes in the Harvard CMT catalog, earthquakes with thrust and strike slip focal mechanisms have significantly lower b values (0.86 and 0.77) than earthquakes with normal faulting mechanisms (1.06). When we separate the ISC catalog into primary events (mainshocks and earthquakes with no aftershocks or foreshocks) and secondary events (aftershocks and foreshocks), we observe that b for secondary events is nearly always significantly higher than b for mainshocks. However, we show that the difference has no physical significance, as it arises simply from the act of choosing mainshocks as the largest earthquake in a foreshock-mainshock-aftershock sequence. When we correct for this systematic effect by comparing the real catalogs to identical catalogs with randomly reassigned magnitudes, we find that b for secondary events in the real catalog is actually lower than expected. Thus among aftershocks large earthquakes are relatively more common than expected, perhaps because the mainshock rupture loads asperities in adjacent regions.

This paper presents a new method for displaying focal mechanisms — plotting them on a ternary graph or ‘triangle diagram’ where the vertices represent normal, thrust, and strike-slip focal mechanisms. This method also provides a natural way for determining the relative proportions of thrust, normal, and strike-slip motion for any particular earthquake focal mechanism.

The role of orbital forcing in glacioeustasy has long presented a major problem regarding the reconciliation of data-oriented seismic sequence stratigraphy with process-oriented dynamic stratigraphy. Dynamic stratigraphy commonly cites orbital forcing periods of approximately 20,000 years (precession), 40,000 years (tilt), and 100,000 years (eccentricity) which are too fine scale to be resolved consistently by seismic sequence stratigraphy. Seismic sequence stratigraphy consistently identifies seemingly eustatic events on a scale of millions of years which have no heretofore easily identifiable causal mechanisms. In this paper, we propose a geologic mechanism and realistic quantitative construct to explain how Antarctic ice volume varies as a function of orbital forcing in the Tertiary. We demonstrate that this mechanism, which has a nonlinear response to long-period modulation of the orbital forcing time series, can produce major glacioeustatic events with quasi-periodicities of the order of 2 m.y. We use a FORTRAN program, STRATA-various, to construct a two-dimensional forward model demonstrating that this proposed mechanism can produce a synthetic sequence stratigraphy which bears strong resemblance to the generalizations of seismic sequence stratigraphy. We acknowledge important achievements of seismic sequence stratigraphy. Nevertheless, we propose that the seismic sequence stratigraphy concepts of long-term and short-term eustatic curves be replaced by independent estimates of tectonoeustasy and glacioeustasy based upon data sets which are, wherever possible, independent of seismic sequence stratigraphy. Likewise, we propose that qualitative generalizations be replaced with explicit forward models as the targets for model/data convergence.

Using single-link cluster analysis, the authors investigate how various properties of aftershock sequences depend on their tectonic regime and focal depth. For International Seismological Centre earthquakes of m{sub b}{ge}4.8, they find that earthquakes deeper than 70 km have the fewest and smallest aftershock sequences. Even after accounting for differences in detectability and maximum magnitude, they find that ridge-transform earthquakes have smaller aftershock sequences that shallow subduction zone earthquakes. Among different subduction zones, they find that zones with high moment release rates possess larger aftershock sequences. Comparing ridge-transform zones, they find those with slower spreading rates possess larger aftershock sequences. By transposing origin times of several different aftershock sequences as if all had main shocks occurring at time zero, they are able to study the properties of aftershock sequences which individually have too few aftershocks to study by other means. Secondary events determined by single-link cluster analysis follow a modified Omori's (power law) decay for time separations of 0.1 day to 20 days from the parent event, with p values ranging from 0.539 {plus minus} 0.022 (intermediate- and deep-focus earthquakes) to 0.928 {plus minus} 0.024 (ridge-transform earthquakes). They find that earthquake foreshocks and multiplets also follow a modified Omori's law. Atmore » greater times from the main shock the decay is steeper than a power law decay, more like an exponential decay. Aftershocks in the Adak catalog (m{sub b}{ge}2.0) show a marked decrease in activity between 40 and 50 km depth. They speculate that the observed differences in number of aftershocks and p values may be caused by variations in fault heterogeneity or in fluid pressures.« less

This paper describes a new method, single-link cluster analysis (SLC), to evaluate percursory quiescence for shallow earthquakes in sixteen subduction zones, using data from the ISC catalog. To define quiescent regions, we divided the catalog into time intervals with a durationT, overlapping byT/2. We considered all earthquakes having magnitudes larger than some magnitudeM min, lying within a specified distance of a great circle which is approximately coincident with the trench near a subduction zone. Within each time interval we connected or ‘linked’ all earthquakes lying within some cutoff distanced of one another. We then projected all these links onto the great circle, and defined a region to be quiescent if it was not covered by the projection of any links. For this study,T was two years,M min wasm b =4.9, and we variedd from 100 to 400 km. We defined an earthquake as ‘following quiescence’ if it occurred within two years following, and within 75 km of a quiescent zone as defined above. The primary conclusion of this study was that earthquakes with surface wave magnitudes 7.2 and greater were about 5–15% more likely to follow quiescence than were the smaller background earthquakes withm b >-4.9. A chi-squared analysis shows that this result is significant at the 99% level. In contrast, earthquakes with surface wave magnitude of 6.7 to 7.1 were no more likely to follow quiescence than were background earthquakes. Of sixteen individual regions, Central America, Japan, and Peru-Chile were the only regions where large earthquakes were more likely to occur following quiescence than were background earthquakes. For a cutoff link length of 300 km, only in Central America was the difference between large earthquakes and background earthquakes significant at the 95% level of significance. For a cutoff link length of 250 km, the significance level exceeded 95% only in Japan. The SLC method is an objective, quantitative method for evaluating large data catalogs, or for monitoring quiescence in regions where quiescence is conjectured to precede large earthquakes.

Astronomers have little confidence in orbital forcing calculations on time scales beyond ten million years. They are confident that the fundamental frequencies of the planetary orbits remain nearly constant on time scales of hundreds of millions of years. Small variations in these frequencies and chaotic behavior of certain resonance relationships, however, can introduce large uncertainty regarding the exact orientation of planetary orbits on time scales of tens to hundreds of millions of years. We introduce a simple trigonometric series expression for eccentricity in order to evaluate effects that arise from imposing specified displacements in orientation of planetary orbits. The resulting changes in the character of the calculated eccentricity time series should be a guide to stratigraphers seeking to relate orbital forcing to stratigraphic time series throughout the Phanerozoic. This geologic calibration of eccentricity time series in relatively old stratigraphic sequences may provide astronomers with clues to further refinement of their calculations.

One suggested explanation for the origin of deviatoric, non-double-couple earthquake sources is that there is slip along curved fault surfaces. Here we compute the moment tensor components for earthquakes along idealized faults which are surfaces of revolution. The calculations show that when the slip direction is around the axis of revolution, the earthquakes may possess purely double-couple sources, or purely compensated linear vector dipole (GLVD) sources, depending on the geometry of the fault and the slip. However, a surprising result is that the CLVD component is always zero for unidirectional slip along surfaces having a suitable degree of symmetry. Thus slip along a large variety of simple and plausible curved fault geometries produces a purely double-couple source.

We have examined teleseismic earthquake locations reported by the International Seismological Centre (ISC) for the Middle America region and selected 220 as the most reliable. These hypocenters and other data are used to delineate the deep structure of the subducted Cocos Plate. The results indicate that the subducted plate consists of three major segments: Segment I extends from the Panama Fracture Zone to the Nicoya Peninsula. The structure of this segment is poorly defined. Segment II is the largest and best-defined segment. This segment consists of two parts, IIA and IIB. Part IIA extends from the Nicoya Peninsula to western Guatemala and is very well defined and continuous in structure. Its strike follows the curvature of the trench and dips at about 60°. Part IIB extends from western Guatemala to Orizaba, Mexico. The dip of this part of the segment decreases slightly toward the northwest, and its strike is more northward than that of the trench. Segment III extends from Orizaba to the Rivera Fracture Zone, and is not well defined due to a lack of earthquake activity beneath about 100 km. Its orientation differs markedly from segment II and strikes somewhat more westward than the trench. Between parts IIA and IIB of segment II the subducted plate seems to be continuous, bending smoothly to accommodate the change in geometry. Local network data from Costa Rica suggest there may be a tear between segments I and II. Between segments II and III there is a gap in the hypocenters which makes it difficult to define the boundary. The change in geometry between these two segments indicates that there may be a tear, and two strike-slip focal mechanisms in the region support this conclusion. We find no convincing evidence supporting the existence of segments smaller than the three described above. If there is smaller-scale segmentation in the shallow part of the subducting plate the plate must still maintain enough continuity to appear continuous at greater depths. There is no evidence for any major tear in the subducted plate associated directly with either the Tehuantepec Ridge or the Orozco Fracture zone. The shallow subduction at the northwestern end of segment II may be related to the bouyancy of the Tehuantepec Ridge. The Cocos Ridge is probably directly responsible for the change in geometry between segments I and II and may even be slowing or stopping subduction in segment I. The structure of the subducted plate in segment II and the changes in the character of volcanism along the arc can be related to the relative motion of the North American and Caribbean Plates. The present geometry of part IIB of segment II is more consistent with the probable configuration of the trench about 7 Ma ago than with the present configuration, indicating that the North American plate is overriding the subduction zone. Appendices 2, 3, and 4 are available with entire article on microfiche. Order from American Geophysical Union, 2000 Florida Avenue, N.W., Washington, DC 20009. Document B84-009; $2.50.

Some earthquake sources are not double couples, i.e, they are not due solely to slip along a simple, planar fault surface. One possible explanation for such sources is that they occur because of source complexity, namely, they consist of simultaneous or time-separated double-couple subevents along different fault surfaces. For intermediate depth earthquakes consisting of only two subevents, we illustrate how non-double-couple earthquakes might occur (1) near the edges of subducting lithospheric slabs; or (2) in the interior of a slab if an earthquake with a downdip P (or T) axis triggers a “bending” subevent along the inside (outside) of the slab. We can choose between an edge, inside-bending or outside-bending interpretation by comparing the downdip direction of the Wadati-Benioff zone and the orientation of the moment tensor axes. While slip along a curved fault surface can produce a non-double-couple source, this can occur only for a limited class of slip and fault geometries. Non-double-couple sources can never be produced by double-couple subevents if (1) the subevents occur along parallel fault planes; (2) the subevents occur on a cylindrical fault surface, with slip due to rotation about the cylinder axis; or (3) the subevents occur on a cylindrical fault surface, with slip parallel to the axis.

Deep earthquakes located in the Tonga-Kermadec region produce exceptionally clear and sharp short-period P, S, PcP, ScP, and ScS phases which are recorded at many stations at distances of less than 60°. The data used in this study are produced by short-period stations located in oceanic-type regions (Fiji and New Caledonia), a mobile continental region (eastern Australia) and a shield region (central Australia). Differential travel-time residuals of the above phases at these stations are investigated to determine the contribution to the differential residuals from: (1) the upper part of the mantle (S-P residuals); (2) the core-to-station portion of the mantle (ScS-ScP residuals); and (3) the hypocenter-to core portion of the mantle (ScP-PcP residuals). The use of differential travel-time residuals considerably reduces near-station effects and effects due to inaccurate determination of the source parameters, and hence the results can be interpreted as due to variations along the propagation paths. The results show that (S-P) residuals from phases traveling along event-to-station paths are about 7 s smaller at the shield station than at the oceanic stations. This correlation with surface tectonic environments is equally strong for the (ScS-ScP) residuals, with the shield/oceanic station difference being about 4 s. Moreover, the data suggest that this correlation between differential residuals and surface tectonic environments is caused by variations in shear velocity within the upper part of the mantle. However, the data cannot uniquely resolve the required depth of these variations within the mantle. For example, if the shear velocity variations extend to a depth of 400 km beneath the recording stations, then the average shear velocity difference between shield- and oceanic-type environments is about 4%. However, if the variations extend only to a depth of 200 km, this difference is more than 8%. (ScP-PcP) and (ScS-PcS) residuals vary from about +1 to about +4 s at the different stations, apparently because of compressional velocity variations in the mantle along the Pc path. If the variation in compressional velocity within the mantle below a depth of about 600 km is about 10% and occurs near the source region, these results suggest that, in the vicinity of deep earthquake zones, variations in compressional velocity extend to a depth of about 1000 km. However, these results can equally be explained by a 1% variation in compressional velocity, evenly distributed along the entire Pc path. An estimate of Q determined from the observed predominant frequency of ScS waves, as recorded at the shield station, suggests that the average 〈Qs〉 of the mantle beneath about 600 km is about 1050 at frequencies of about 1 Hz.

Recent animal sensory literature has been analyzed to evaluate the plausibility of reports of unusual animal behavior prior to earthquakes. Some species of fish possess remarkable sensitivity to pressure waves with frequencies below 50 Hz, which would enable them to sense earthquakes at least 1 to 3 Richter Magnitudes smaller than those detectable by human beings. Fish also have organs which can detect slight movements of water, although the available experimental data do not allow us to determine quantitatively whether they are sensing accelerations of water or variations in water displacement. Thus observations of unusual behavior of fish before earthquakes may be explained if the fish are responding to small foreshocks.

We describe a new statistical method for identifying pairs or groups of related events in sequences which resemble, but are not identical to, a Poisson process. For a particular earthquake, we form the ratio r(Nb,Na) of the relative origin times of the Na th subsequent event and the Nb th previous event. We then find the probability that such a ratio could occur if the sequence were a Poisson process. Presumably, sequences where the ratio is too small to be probable contain related events, and the subsequent events are aftershocks. Since the method requires knowledge of the origin times of only a few preceding and subsequent events, it is more powerful than methods which require knowledge of the mean activity rate of the Poisson process. Using this ratios method, we searched the ISC catalog for aftershocks of earthquakes with focal depths exceeding 70 km. From the world as a whole, events with at least one aftershock can be found in all depth ranges, including 250 km to 450 km. However, the incidence of aftershocks is significantly lower for events between 100 and 450 km than for shallower or deeper events.

The maximum depth of seismic activity is a fundamental observation providing a constraint on models of mantle dynamics. Although most recent investigations of mantle seismicity and dynamics state that seismicity extends to “about 700 km,” focal depths of 720 km have been reported for large earthquakes, as well as depths greater than 800 km for small events. We have examined focal depths of events in several catalogs and conclude that (1) the deepest reliable focal depths are at about 670–680 km, and events with focal depths beneath 600 km occur in several widely separated geographic regions, (2) observations supporting focal depths exceeding 680 km are usually few, poor, or inconsistent, (3) the reduction of seismic activity beneath about 650 km is quite abrupt. However, we cannot determine whether activity stops completely at 670–680 km, or the maximum size of events decreases gradually beneath 650 km, with magnitudes of 6.5–7.0 occasionally occurring at 650 km, magnitudes of 5.5–6.0 at 680 km, 4.0–4.5 at 695 km, etc. Several models can explain the absence of seismic activity in the lower mantle. Two models are consistent with the abrupt termination of activity at about 650–680 km in widely separated areas: a barrier which resists penetration by the subducting lithosphere or a phase transition which modifies the properties of the subducted material as it goes into the lower mantle.

To determine whether reports of unusual animal behavior before earthquakes are plausible, we have compared these reports with recent laboratory studies of animal sensory thresholds. Our major conclusion is that some animals are much more capable than humans of perceiving certain kinds of geophysical stimuli which may precede earthquakes. These geophysical stimuli are seismic or acoustic waves at low frequency (below 50 Hz), electric field changes, and olfactory stimuli. For example, recent studies suggest that some birds and fish are more sensitive than humans to sounds with frequencies below 40 Hz, and many animals are exceptionally good at perceiving low-frequency vibrations through their skin. Certain fish are sensitive to electric field changes as small as 10−5 V/m, and some laboratory mammals also respond to significantly weaker fields than humans. For these electric and acoustic stimuli the reported levels of geophysical precursors are within the reported perceptible range of some animals which show unusual behavior prior to earthquakes. In addition, stimuli caused by the release of gases from small cracks may well be perceived by some animals before earthquakes. Recent research has confirmed the remarkable olfactory sensitivity of some animal species, but no quantitative comparisons with geochemical precursors can be made yet. We find no evidence that magnetic field precursors or precursory high-frequency (above 10 kHz) sounds are the cause of unusual animal behavior before earthquakes. Although some animals are highly sensitive to these stimuli, observed magnetic precursors tend to be of low amplitude amid relatively high noise levels, and high-frequency sound is severely attenuated within a short distance from the earthquake hypocenter. Knowledge of animal sensory capabilities may suggest an instrumental strategy for detecting earthquake precursors. We recommend further geophysical research in seismic areas to measure (1) seismic waves in the frequency range 10–50 Hz, (2) earthquake-related electric field changes and air ionization, and (3) coseismic or preseismic release of trapped gases other than radon. Further biological research can refine our understanding of (1) responses to sounds and vibrations with frequencies below 50 Hz, (2) the effects of electrical and electrostatic changes on animal behavior, (3) behavioral responses to odors, and (4) sensory capabilities of common domestic animals like horses, dogs, and chickens, which are seldom studied but commonly mentioned in the preearthquake reports.