Lg Waves Research Articles

An inversion of site response and Lg attenuation using Lg waveform

Based on spectral ratio method, a joint inversion method was used to obtain parameters of Lg wave attenuation and site response. The inversion method allows simple and direct (two-parameter) determination of Lg wave attenuation and site response from sparse spectral data, which are not affected by radiation pattern factor and different response of same instrument after geometrical spreading. The method was used successfully for estimating site response of stations of Zhejiang Seismic Network and measuring Lg wave attenuation. The study is based on 20 earthquakes occurred in northeast of Taiwan with magnitude MS5.0–6.7 and 960 seismic wave records from 16 stations in Zhejiang area from 2002 to 2005. The parameters of site response and Lg attenuation were calculated with a frequency interval of 0.2 Hz in the range of 0.5 Hz to 10 Hz. Lg wave attenuation coefficient corresponding to U-D, E-W and N-S components are γ(f)=0.00175f0.43485, γ(f)=0.00145f0.48467 and γ(f)=0.0021f0.41241, respectively. It is found that the site response is component-independent. It is also found that the site response of QIY station is significant above the frequency of 1.5 Hz, and that the site response of NIB station is low for most frequency

Use of Time-Deformation Methods to Discriminate between Earthquakes and Explosions on the Basis of Lg Alone

In this article we introduce the use of time-transformation methods for analyzing signals with time-varying frequencies. We show that these techniques can be usefully applied to the Lg waves of earthquakes and explosions for purposes of deriving discriminants based on the Lg wave alone. Preliminary examination of the Lg waves from 10 earthquakes and 10 explosions recorded at the Mina, Nevada (mnv), station shows that frequencies are varying (decreasing) more systematically with time for explosions than for earthquakes. Use of a proposed discriminant produced excellent discrimination between the earthquakes and explosions analyzed.

Q of Lg Waves in the Central Mexican Volcanic Belt

From seismograms of nine shallow, coastal earthquakes recorded at a pair of broadband stations, we estimate Q of Lg waves in the part of central Mexican Volcanic Belt (mvb) that includes the Valley of Mexico. The two stations straddle the central mvb and are located on Cretaceous limestone. A weighted least-square fit to the Q −1 ( f ) data in the frequency range 0.25 to 8 Hz yields Q ( f ) = 98 f 0.72 . This estimate of Q is lower than the corresponding Q in the forearc region that is given by Q ( f ) = 273 f 0.66 . Note that our estimate of Q ( f ) corresponds to a 200-km-wide zone of the mvb. The result of this study sheds light on the characteristics of seismic waves as they traverse through the mvb where they undergo dramatic amplification in the Valley of Mexico. It also provides one of the critical elements needed in the estimation of expected ground motions at sites to the north of the mvb from future coastal earthquakes. The lower Q of Lg waves in the mvb as compared with the forearc region seems correlated with lower resistivity reported in the mvb relative to the forearc region.

An improved UK local magnitude scale from analysis of shear andLg-wave amplitudes

The amplitudes of shear and Lg waves recorded at UK seismograph stations from local earthquakes in the range 0–600 km have been subjected to an analysis of variance, allowing separation of the effects of earthquake size, distance and station corrections for each recording site. The propagation paths sample mainly the central to western part of the UK, with good north–south coverage. The analysis of 385 amplitude readings at 28 stations from 40 earthquakes showed that the effects of both distance and station (site) corrections were statistically significant. Tables of corrections for both distance and station effects have been derived to allow local magnitude ML to be determined more accurately from horizontal and vertical component records. One set of tables allows the estimation of an ML which is consistent with the original Richter definition of ML, with a standard deviation which is smaller than that produced by the theoretical attenuation curve, defined according to attenuation in Southern California, which has customarily been used to calculate ML in the UK. The reduction in standard deviation is mainly due to the incorporation of station terms to correct for station effects, since the theoretical and observed variations of attenuation with distance are similar. This similarity implies that Southern California and the UK show a similar variation of distance-dependent attenuation for Lg waves, which was not expected in view of the differences in geology and tectonics. Another set of tables is provided to calculate an ML which is consistent with the body wave magnitude mb determined by the International Data Centre from station records of the Comprehensive Test-Ban Treaty Organization’s global monitoring network. The decay with distance of the predominantly Lg-wave amplitude values gives a value of 440 ± 50 for Lg Q at a frequency of 3 Hz, in agreement with estimates obtained from spectral displacement amplitudes of Lg waves in the UK.

Seismic-Wave Attenuation and Source Excitation in La Paz-Los Cabos, Baja California Sur, Mexico

We present results from a regional study of seismic-wave attenuation and source excitation from small-magnitude earthquakes recorded at distances from 6 to 180 km in the La Paz–Los Cabos region, at the south end of the Baja California Peninsula. Data were recorded using 32 strong-motion seismic stations from the La Paz network (lap). A least-squares regression separating the excitation, site, and propagation effects was carried out. We performed the analyses in the time and frequency domains, and we compared these results with results from a coda- normalization method. The propagation term was parameterized to represent a geometrical spreading function and a frequency-dependent Q ( f ) at a reference distance of 40 km. We estimated the regional attenuation by measuring the maximum amplitude of the S or Lg waves as a function of frequency, defining a continuous piecewise propagation term, D ( r , f ), after separating the excitation and site terms. Our results show that the attenuation is lower than that of central or northern Mexico. Recorded data were of remarkably good quality despite the fact that the strong- motion network recorded only small earthquakes. Our best model is that of a quality factor of Q ( f ) = 380 f 0.10 with a complex geometrical spreading function.

Tomographic regionalization of crustal Lg Q in eastern Eurasia

We measure Q of crustal Lg waves in eastern Eurasia using the reliable two‐station methods. More than 5,000 spectral ratios are collected over 594 interstation paths, and are used to determine values of Lg Q0 and η (Q at 1 Hz and its frequency dependence) over these paths. These values are used to derive tomographic models of laterally varying Q0 and η with resolutions between 4 and 10°. The Q0 model contains values that vary between 100 and 900. Q0 are the lowest in the Tibetan plateau, increase to moderate levels towards the east and north, and reach maxima in the Siberian and eastern Europe Cratons. Q0 values correlate well with regional tectonics. Most η values are between −0.14 and 0.50, with a mean of 0.18. These values are lower than those obtained previously using either an Lg coda Q method, or a method of simultaneous inversion of source spectra and Q.

Phase composition of regional seismic waves from underground nuclear explosions

The mechanism of regional shear wave excitation by underground nuclear explosions (UNE), a long‐standing issue in nuclear seismology, is investigated by studying the phase composition of regional waves from clustered UNE. Regional seismic waves from 67 UNE at the Balapan test site of Kazakhstan, recorded at station Borovoye, are analyzed with a source array technique under the reciprocity theorem. This analysis allows one to obtain the phase velocity composition of the plane waves leaving the source region. Source locations and the original times are obtained from geodetic measurements, a scaling law between event magnitude and depth, and a calibrated Pn travel time curve. Between frequencies of 0.5 and 3.0 Hz the Pn slowness power spectra are concentrated at a phase velocity (vh) of 8.0 km/s. The expected Sn window contains two dominant phases: a scattered P wave with a vh of around 7.1 km/s and a mantle shear wave with a vh of 4.8 km/s. The Lg waves are coherent between 0.5 and 2.0 Hz, with a dominant vh of 4.2 km/s, which is typical in SmS‐type waves. The Rg wave at frequencies between 0.5 and 0.8 Hz is dominantly composed of a coherent fundamental mode Rayleigh wave with a vh of 3.0 km/s. At higher frequencies (>0.8 Hz) this coherent Rayleigh wave is not observed in the Rg window because of attenuation during wave propagation. The slower vh of Rg and faster vh of Lg, at which the strong coherent peaks in slowness power spectra are observed, suggest that the dominant components of these two waves are different and they are originated differently in the source region. The proposed Rg‐to‐S scattering does not appear to be a dominant mechanism for Lg excitation in the Balapan test site. It is also found that the strength of shear waves contained in the expected Sn window varies with local geology much more than the strength of Lg does. Since the Sn is enriched in high‐frequency (>1 Hz) content as compared to the Lg, this observation suggests that the local geology influences shear wave excitation at high frequencies more than at lower frequencies.

A transportable scale for central Europe and implications for low-magnitude discrimination

SUMMARY We have extended the mb(Lg) method of Nuttli using root-mean-square (rms) amplitudes corrected for noise and a Δ−1 dependence for geometrical spreading. Lg waves recorded on the German Regional Seismic Network (GRSN) for earthquakes in south-central Europe were used to develop an mb(Lg) formula requiring a new calibration constant Crms to keep rms mb(Lg) on the same baseline as Nuttli's traditional formula based on 3rd-peak amplitudes. GRSN stations had to be calibrated for site terms and for Lg attenuation. Lateral variations in LgQ appear to be significant across the study area, and a regional Q model consisting of constant-Q partitions north, south and in the central Alps was developed using measurements based on interstation and two-event, single-station methods. When plotted against surface wave estimates of Mw, rms mb(Lg) measurements in central Europe are found to be consistent with Mw–mb(Lg) relationships for north America and southern Asia, thus supporting the transportability of our mb(Lg) formula. Frequency–wavenumber processing of Grafenberg Array data enabled us to extract Rayleigh waves for small events, and regional Ms were measured using the Marshall and Basham formula. Our Ms–mb(Lg) relationship extends to Ms 2.5 and agrees well with observations in other regions including the western United States. The discrimination potential of Ms–mb(Lg) observations was examined under realistic monitoring conditions, where path corrections were inferred from earthquake data and applied uniformly to natural sources and explosions. Under these conditions, mb(Pn, P) are greater than mb(Lg) for large NTS explosions; however, Ms–mb scaling slopes are steeper for P waves than they are for Lg, and Ms−mb observations for NTS explosions converge near mb 4. Thus, allowing for measurement errors and additional uncertainty in mb(Pn) due to regional bias, there is little difference in the discrimination potential for Pn and Lg waves at small magnitudes. As such, a regional Ms−mb discriminant based on Lg might be preferred owing to the better detectability of Lg waves for small earthquakes. These results need to be confirmed for explosions at other test sites. Compared to teleseismic experience, regional Ms−mb observations extend the discrimination capability to lower magnitudes by at least one Ms unit.

The effect of spall on Lg waves

The generation mechanism of Lg wave from underground nuclear explosion is still not clear at present. The general viewpoint is that the S wave generated by the near-source scattering of explosion-generated Rg appears to be the primary contributor to the low-frequency Lg (0.2–2.0 Hz) from nuclear explosions. The viewpoint is supported by the analysis of regional data from several Yucca Flats, NTS explosions by Patton and Taylor (1995), who further indicated that the prominent low-frequency spectral null in Lg is due to Rg from a compensated linear vector dipole (CLVD) source. In the paper, the data from Kazakstan Test Sites are investigated by a spectral ratio method. We have found that the spectral ratio of Lg waves is characterized by a spectral scalloping and a pronounced null, and the spectral null does not shift with spall dwell times, showing a strong dependence on shot depth and a very good agreement with those expected from Rg due to a CLVD source.

The Excitation of Lg Waves by Underground Nuclear Explosions

Near-source scattering of explosion-generated Rg wave into S wave appears to be the primary contribution to the low-frequency Lg wave in nuclear explosions, this mechanism has been supported by analysis of regional data from several Yucca Flats, NTS explosions by Patton and Taylor (1995), who further indicated that the prominent low-frequency spectral null in Lg wave is due to Rg wave from a compensated linear vector dipole (CLVD) source. In this study, the mechanism of the excitation of Lg wave in explosions in layered earth structures representative of the East Kazakh test site is further analyzed by using theoretical seismograms, and the result shows that the Lg spectra indicate prominent nulls showing strong dependence on focal depth and in remarkably good agreement with those expected from Rg wave due to a CLVD source.

Lateral variations of crustal seismic attenuation along the INDEPTH profiles in Tibet from Lg Q inversion

Fourier spectra of crustal guided Lg waves are collected to study seismic attenuation, or Q, along the INDEPTH profiles. The 1 Hz Lg Q (Q0) values are measured between many pairs of two stations and input to a back projection algorithm to obtain laterally varying Q0 models. Along the INDEPTH III profile in central Tibet, the Q0 model consists of low and nearly constant values of ∼90. These low values are consistent with abnormally high temperature as well as partial melts in the crust. Along the INDEPTH II profile in southern Tibet the Q0 values increase southward. They are extremely low (∼60) over the northernmost segment but increase by a factor of 2 over a distance of ∼100 km between the Indus‐Yalong Suture and the Kangmar Dome. Farther south and into the high Himalayas, Q0 values are higher than 300. Regional Rayleigh waves observed along the northern INDEPTH II profile are used to infer a low‐velocity, low‐Q layer at midcrustal depths (between ∼15 and 30 km) in southern Tibet. The aqueous fluid trapped in the upper crust and a midcrust partial melting zone, associated with the underthrusting of the Indian lithosphere, are the likely causes of the high Lg attenuation in southern Tibet. Low Q, rather then a systematic Lg conversion along a dipping Moho, is the main cause of the previously observed Lg blockage over paths crossing the southern Tibetan boundary.

Q of the Indian Shield

We analyze spectral attenuation of Lg waves to determine Q of the Indian shield. The dataset consists of four earthquakes recorded in the distance range of 240-2400 km. The new estimate, Q(f ) 800f 0.42 (0.1 Hz f 20 Hz), is based on a larger dataset than a previous estimate, Q(f ) 508f 0.48 (1 Hz f 20 Hz). The new Q(f ) of the Indian shield is comparable to the Q in regions of eastern North America where its value is relatively high, for example, the Adirondack Mountains. It is also in agreement with previously reported Lg coda Q at f 1 Hz for the Indian subcontinent. The revised Q(f ) has important implications for the estimation of ground motions in the Indian shield region using the stochastic method since the recorded data, presently, come mostly from large distances.

Lg-Wave Cross Correlation and Double-Difference Location: Application to the 1999 Xiuyan, China, Sequence

A surprising discovery has been made that in some cases the complex, highly scattered Lg wave is found to be similar for clusters of events. We analyze in detail a subset of 28 out of 90 events from the 1999 Xiuyan sequence. Cross correlations provide highly accurate differential travel-time measurements. Their error estimated from the internal consistency is about 7 msec. These travel-time differences are then inverted by the double-difference technique to obtain epicenter estimates that have location precision on the order of 150 m. The locations are computed with waveform data from four to five regional stations 500 to 1000 km away. The epicenter estimates are not substantially affected by the sparseness of stations or large azimuthal gaps. Comparison with a surface trace a few kilometers away and location estimates based on much more dense networks led us to conclude that the absolute positions are accurate to the 5-km level. Regional event locations must often be based on a small number of phases and stations due to weak signal-to-noise ratios and sparse station coverage. This is especially true for monitoring work that seeks to locate smaller magnitude seismic events with a handful of regional stations. Two primary advantages of using Lg for detection and location are that it is commonly the largest amplitude regional wave (enabling detection of smaller events) and it propagates more slowly than P waves or Sn (resulting in smaller uncertainty in distance, for a given uncertainty in travel time).

Separation of intrinsic absorption and scattering attenuation fromLgcoda decay in central France using acoustic radiative transfer theory

Radiative transfer theory is applied to the measurement of average values of the transport mean free path and absorption length (l a , related to the intrinsic quality factor) in the lithosphere beneath France by comparingsimulated and observed coda envelopes of regional records. The lithosphere is modelled as a flat layer representing the crust overlying a half-space representing the mantle, with different velocities and scattering properties. We infer the average values of I a (assumed to be constant over the whole medium) and transport mean free paths in the crust (l * c ) and mantle (l * m ) by minimizing a misfit function between observed and modelled Lg coda envelopes. This comparison is conducted in two different time windows, the whole Lg coda (group velocity smaller than 2.6 km s - 1 ) and the early coda (v g < 2.6 km s - 1 and lapse time smaller than 300 s), when coda envelopes have decay rates that depend on distance. In the early coda, a single set of parameters l a and l* c minimizes the misfit function. Using the whole Lg coda time window we find that many models fit the data equally well. We explain this difference by the common decay rate of envelopes independent of distance over most of the coda window. The measurement of the attenuation of the coherent Lg wave complements these results to constrain l a and l* c . We find that a model with l* c ∼ 250 km and l a ∼ 150 km explains the attenuation of both the direct Lg wave train and the entire Lg coda with both time and distance at 3 Hz. However, the trade-off between the two parameters is too strong to conclude reliably on the dominance of intrinsic on scattering attenuation.

A quantitative study of the energy release in the aftershocks of the Bhuj earthquake, 2001, India, using Lg phase

The devastating earthquake on 26 January 2001 at Bhuj, India, resulted in large-scale death and destruction of properties of several million US dollars. The moment magnitude of the earthquake was 7.7 and its maximum focal intensity exceeded X in MM scale. The rate of aftershocks of this earthquake, recorded at Gauribidanur seismic array station (GBA), shows a monotonic decay with time superposed with oscillations. For the Indian continent the Lg phase is a prominent arrival at regional distances. The estimate of Lg amplitude is obtained by optimally fitting the Lg wave train to a exponential decay curve. The logarithm of these amplitudes and logarithm of root mean square (rms) value of actual amplitudes of the Lg are calibrated with USGS m b to create a local m bLg magnitude scale. The energy released from these aftershocks is calculated from the rms value of Lg phase. The plot of cumulative energy release with time follows the power law of the form t p , superposed with oscillations. The exponent of the power law, p, is estimated both by a time-window scanning method and by an interpolation method. The value of p is 0.434 for time-window scanning method and 0.432 for the interpolation method. The predominant periods found in the oscillatory part of the cumulative energy, obtained by differencing the observed from the power law fit, are 10.6, 7.9, 5.4, 4.6 and 3.5 h for time-window scanning method. The corresponding periods for interpolation method are 13.4, 11.5, 7.4, 4.2, 3.5, 2.6 and 2.4 h.

Visualization of 3D Wave Propagation from the 2000 Tottori-ken Seibu, Japan, Earthquake: Observation and Numerical Simulation

The dense networks of strong-ground-motion instruments in Japan (K- NET and KiK-net) provide a means of direct visualization of regional wave propa- gation during large earthquakes. For the 2000 Tottori-ken Seibu earthquake (Mw 6.6) in western Japan, snapshots of ground motion, derived directly from interpolation of a large number of array observations, demonstrate clearly the nature of the source radiation and the character of the seismic wave field propagating to regional dis- tances. In western Japan the wave field from the earthquake is characterized, in most parts, by the dominance of high-frequency (0.2-5 Hz) Lg waves on three-component acceleration records and longer-period (T 10 sec) fundamental-mode Love waves in tangential displacement. The presence of strong lateral variations in the crust and upper-mantle structures, such as the low-velocity superficial layer and the high- velocity Philippine Sea plate with shallow subduction into the mantle, impose sig- nificant modifications on the regional wave field. Further insight into the nature of the seismic wave field is gained by comparison of the observations with numerical simulation for a 3D model, including sedimentary basins by an embedded submesh. A multigrid, parallel computation using a hybrid Pseudospectral/Finite Difference Method allows the inclusion of a realistic model of the source process for the 2000 Tottori-Ken Seibu earthquake. There is good agreement in the dominant features of the regional wave field propagating through the complex structure of western Japan. However, the differences between the observations and the computer simulations indicates the need for further refinement of the source and structural models. The inclusion of amplification effects for the population centers in sedimentary basins means that the modeling procedure is already suitable for estimating the main pattern of ground motion for future earthquake scenarios.

Earthquake Ground-Motion Scaling in Central Mexico between 0.7 and 7 Hz

We present results from a regional study of ground-motion scaling parameters in central Mexico using data from a short-period vertical-component network and two broadband stations within the valley of Mexico. A total of 1220 waveforms recorded between 1994 and 2001 are used for the analysis. A damped least-squares regression using a simple model to separate the excitation, site, and propagation effects for the peak velocities in selected narrow-bandpass frequencies was carried out. The propagation term was parameterized to define a piecewise continuous geometrical spreading function, a frequency-dependent Q ( f ), and a distance-dependent duration that are consistent with the random vibration theory. We measured the average attenuation of S and Lg waves. A final model with a quality factor of Q ( f ) = 180 f 0.66 and a geometrical spreading of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \[g(r)=\begin{array}{ll}r^{-1}&0{<}r{\leq}100\\r^{-0.2}&100{<}r{\leq}150\\r^{-0.5}&150{<}r{\leq}300\end{array}\] \end{document} is used for the parameterization of the ground-motion scaling. We find that this region is characterized by a rapid decay of ground-motion amplitude with distance, similar to that in other tectonically active regions. Manuscript received 23 October 2001.

Estimation of Ground Motion for Bhuj (26 January 2001; Mw 7.6) and for Future Earthquakes in India

Only five moderate and large earthquakes (Mw 5.7) in India—three in the Indian shield region and two in the Himalayan arc region—have given rise to multiple strong ground-motion recordings. Near-source data are available for only two of these events. The Bhuj earthquake (Mw 7.6), which occurred in the shield region, gave rise to useful recordings at distances exceeding 550 km. Because of the scarcity of the data, we use the stochastic method to estimate ground motions. We assume that (1) S waves dominate at R 100 km and Lg waves at R 100 km, (2) Q 508f 0.48 is valid for the Indian shield as well as the Himalayan arc region, (3) the effective duration is given by fc 1 0.05R, where fc is the corner frequency, and R is the hypocentral distance in kilometer, and (4) the acceleration spectra are sharply cut off beyond 35 Hz. We use two finite-source stochastic models. One is an approximate model that reduces to the x 2 -source model at distances greater that about twice the source dimension. This model has the advantage that the ground motion is controlled by the familiar stress parameter, Dr. In the other finite-source model, which is more reliable for near-source ground-motion estimation, the high- frequency radiation is controlled by the strength factor, sfact, a quantity that is phys- ically related to the maximum slip rate on the fault. We estimate Dr needed to fit the observed Amax and Vmax data of each earthquake (which are mostly in the far field). The corresponding sfact is obtained by requiring that the predicted curves from the two models match each other in the far field up to a distance of about 500 km. The results show: (1) The Dr that explains Amax data for shield events may be a function of depth, increasing from 50 bars at 10 km to 400 bars at 36 km. The corresponding sfact values range from 1.0-2.0. The Dr values for the two Himalayan arc events are 75 and 150 bars (sfact 1.0 and 1.4). (2) The Dr required to explain Vmax data is, roughly, half the corresponding value for Amax, while the same sfact explains both sets of data. (3) The available far-field Amax and Vmax data for the Bhuj mainshock are well explained by Dr 200 and 100 bars, respec- tively, or, equivalently, by sfact 1.4. The predicted Amax and Vmax in the epi- central region of this earthquake are 0.80 to 0.95 g and 40 to 55 cm/sec, respectively.

LgwaveQtomography in Central America

SUMMARY The lateral variation of Lg wave spectral attenuation in Central America was studied through tomographic inversion at various frequencies between 0.5 and 5 Hz, separately for each frequency. The input data set consisted of 558 travel paths recorded on short-period and broadband seismic stations in the region. The size and quality of the data set were sufficient to resolve significant lateral variation in the Lg wave quality factor Q Lg. The average dependence of Q Lg with frequency was found to be Q Lg( f ) = 182 f 0.84 , corresponding to high attenuation. Low Q Lg values in the Nicaraguan Depression presented the most significant variation from the average. The strong attenuation of Lg waves in the Nicaraguan depression was also observed in the visual analysis of the seismograms. This observation is possibly explained by the nearsurface low-velocity layers in the depression. Low Q Lg values were also found along the chain of volcanos due to increased scattering and partial melting beneath the volcanic belt. This study is the first attempt to determine Q Lg for Central America and to provide knowledge on attenuation of Lg waves needed for the prediction of ground motion during future earthquakes.

QLg tomography in Colombia

The crustal attenuation of Lg waves in Colombia was estimated and analyzed using local seismological data from the National Seismological Network of Colombia (RSNC). The selected dataset comprises 510 crustal earthquakes with a total of 2928 ray paths. This large dataset allowed us to invert for a regional average of Q Lg, tomographic images for Colombia, and simultaneously, source size for each event and site term at the seismological stations. The computed regional average for Q Lg in the frequency band 0.5–5.0 Hz was found to agree with the previously reported values in neighboring regions. In order to resolve the lateral variation in the attenuation of Lg waves in the crust, independent tomographic inversions for 26 frequencies between 0.5 and 5.0 Hz were conducted. The resulting maps confirm that heterogeneities in the crust exist and that they are related to the large-scale tectonic features in the country. The highest attenuation in the region is linked to the presence of active volcanic arcs, where the crust is weakened by partial melting. Relatively high attenuation is also found in zones where the crust is composed of accreted oceanic rocks and in regions with near-surface low-velocity sedimentary layers.