Spectral Amplitude Ratios Research Articles

Seismic discrimination with artificial neural networks: Preliminary results with regional spectral data

An application of artificial neural networks (ANN) for discrimination between natural earthquakes and underground nuclear explosions has been studied using distance corrected spectral data of regional seismic phases. Pn , Pg , and Lg spectra have been analyzed from 83 western U.S. earthquakes and 87 Nevada Test Site explosions recorded at the four broadband seismic stations operated by Lawrence Livermore National Laboratory. Distance corrections are applied to the raw spectra using existing frequency-dependent Q models for the Basin and Range. The spectra are sampled logarithmically at 41 points between 0.1 and 10 Hz for each phase and checked for adequate signal-to-noise ratios ( S/N > 2). The ANN was implemented on a SUN 4/110 workstation using a backpropagation-feedforward architecture. We find that, using even simple ANN architectures (82 input units, 1 hidden unit, and 2 output units), powerful discrimination systems can be designed. In order to regionalize the data characteristics, a separate neural network was assigned to each station. For this data set, the rate of correct recognition for untrained data is over 93 per cent for both earthquakes and explosions at any single station. Using a majority voting scheme with a network of four stations, the rate of correct recognition is over 97 per cent. Although the performance of the ANN is similar to that of the Fisher linear discriminant, the ANN exhibits a number of computational advantages over the conventional method. Finally, examination of the network weights suggests that, in addition to spectral shape, a criterion that the ANN utilized to discriminate between the two populations was the Lg/Pg spectral amplitude ratios.

Seismic ground-response studies in Olympia, Washington, and vicinity

AbstractRelative seismic ground-response characteristics in the cities of Olympia, Lacey, and Tumwater, Washington, were determined from analysis of instrumentally recorded ground motion induced by blasts at an open-pit coal mine near Centralia, Washington. A ground-response function (GRF), defined as the ratio of Fourier spectral amplitudes at an alluvium site to spectral amplitudes on hard rock, is a measure of amplification of seismic waves by localized site conditions. GRF values in three frequency bands (0.5 to 1.0 Hz, 1.0 to 2.0 Hz, and 2.0 to 4.0 Hz) were compared with observed Modified Mercalli (MM) intensities from the 29 April 1965, Puget Sound earthquake and with mapped surficial geologic units. Typically, the GRF values relate well with the surficial geological units.In addition, MM intensities within the V to VII range appear to be directly related to the frequencies within the 0.5 to 4.0 Hz bandwidth such that MM V intensity sites had a lower GRF value in the 2.0 to 4.0 Hz bandwidth as compared to the 0.5 to 2.0 Hz bandwidth, and the MM VII intensity sites had higher GRF values in the 2.0 to 4.0 Hz bandwidth as compared to the 0.5 to 2.0 Hz bandwidth. The set of GRF values determined for the city of Olympia and its vicinity should be useful in formulating a theoretical relative ground-response model for the southern Puget Sound area.

Strain and frequency-dependent attenuation estimates in salt based on SALMON and STERLING near-field recordings

Abstract The amplitude and spectra of teleseismic and regional seismic P waves depend on the material properties in the source region. The effects of strain and frequency-dependent attenuation are therefore important for understanding the source characteristics of coupled versus decoupled explosions. Near-field velocity and acceleration data from the SALMON and STERLING explosions in salt were analyzed for strain and frequency-dependent attenuation. Attenuation, parameterized as 1/Q, was estimated for various source-receiver distance ranges and frequency bandwidths using two methods: mean spectral ratio slope and average amplitude spectral ratio. Results from SALMON indicate that (1) in the frequency range of 1 to 25 Hz, Q is about 5 to 10; (2) Q increases with source-receiver distance, suggesting lower Q for larger strain levels; and (3) Q in the frequency range of 25 to 50 Hz is substantially higher than in the frequency range of 1 to 25 Hz, or Q increases with frequency. Analysis of data from STERLING indicates that Q is frequency independent and about one order of magnitude larger than for SALMON. It appears therefore that the near-source attenuation was strain and frequency dependent and hence nonlinear for SALMON out to a range of at least 600 m.

Slab Decoupling in the Tonga Arc: The June 22, 1977, Earthquake

The June 22, 1977, earthquake in the southern Tonga arc is a large normal faulting event with downdip rupture propagation on a steeply dipping fault plane. We use a standard least squares inversion of P and SH waves for the source time function to determine a source consisting of two subevents located at depths of 40–55 and 105–125 km and a total time duration of 50 s. The ratios of the spectral amplitudes of the radial modes 0S0, 1S0 and 2S0 independently require a centroid depth of at least 100 km at their lower frequencies. The body wave analysis suggests a moment of 1.04 × 1028, and the 0S0 spectral amplitude suggests a value of 1.8–2.0 × 1028 dyn cm. The emerging picture is that of a complete rupture of the sinking lithospheric slab, starting from 40 km depth and propagating down to 125 km. This indicates that the decoupling of the overriding lithosphere from the slab in this area of the Tonga arc takes place by gravitational pull.

Reflections of P'P' from upper mantle discontinuities beneath the Mid-Atlantic Ridge

SUMMARY Precursors to P'P’ observed in Japan from a large deep earthquake (mb= 6.4) in the Fiji Islands are analysed to investigate reflection properties of upper mantle discontinuities beneath an area close to the Mid-Atlantic Ridge (MAR) in the equatorial Atlantic Ocean. Traveltime analysis indicates that the precursors are interpreted as underside reflections of P'P’ from discontinuities at depths of 655 and 415 km in the mantle. The latter reflection is much weaker and more intermittent than the former. Slant-stacking and beam-forming are attempted to measure the slowness and incidence azimuth of P'P’ and its precursors. The obtained slownesses are 3.4 and 2.8 s deg-1 for P'P’ and P'P', respectively. Little deviations of the ray incidence off the diametral azimuth are found for both phases. Amplitude ratio P'dP'/P'P’ (here d= 655 or 415) is studied as a function of the wave period and the distance of reflection point of P'P’ to the axis of MAR. The overall trend of the spectral amplitude ratio for P'655P’ is consistent with the theoretical value for the 670 km discontinuity of the PREM model (Dziewonski & Anderson 1981). P'415P’ signals were very weak and it was difficult to obtain an unbiased estimate of P'415P'/P'P'. The amplitude ratios P'655P'/P'P’ of filtered seismograms with a passband of 0.5-4.0 s have a mean value of 0.13 and a standard deviation of 0.03. This amplitude ratio is in good agreement with those of previous studies on the reflections from the 650-km discontinuity beneath Antarctica and surrounding seas. When the reflection points of P'P’ are calculated under the assumption of a spherically symmetric earth, the observed P'655P'/P'P’ seems to decrease as the points become close to the axis of MAR. Thus it could be inferred that some changes in the reflectivity of the discontinuity might occur at or near the spreading center. Another interpretation may also be possible. The low velocity beneath the MAR region would cause focussing of P'P’ rays.

Q measurements from compressional seismic waves in unconsolidated sediments

Several processing methods in both time and frequency domains have been used to compute the in‐situ specific dissipation function 1/Q, or its inverse, the quality factor Q, in water‐saturated unconsolidated sediments. These methods are based on measurements of spectral amplitude ratio, peak‐to‐peak and first‐peak amplitude ratio, rise time, pulse broadening, and the Futterman causal attenuation operator of an attenuating signal. Compressional seismic waves were generated from explosive sources ranging in size from 1 to 64 mg of silver azide and recorded at a depth of 7.62 m below the surface near Wendover, Utah. The medium consisted of silty sandy clays with mean grain size diameter of 7.48 μm and water saturation of 70 percent. The hydrophone receivers were spirally distributed at distances ranging from 25 to 200 m from the source. An average Q value of 26 was obtained using analysis of spectral amplitude ratios over the frequency range 450–725 Hz for the five different sizes of explosive sources. Measurements of peak amplitude ratio produced an average Q value of 123. The computed Q value from the rise‐time method is found to have a wide range of Q values, from 50 to 207. The Q values determined from the pulse‐broadening technique ranged from 25 to 158 for quarter‐cycle measurements and from 26 to 114 for half‐cycle measurements. The Futterman causal attenuation operator yields high Q values, on the order of 200 to 300. The wide variation of the computed Q values indicates that different computational techniques can result in different Q values for the same type of materials. Results show that the rise‐time and pulse‐broadening methods are probably source‐dependent. This makes the spectral amplitude‐ratio method preferable to these two methods since it is independent of the source. A correct geometrical spreading factor appears to be sufficient to account for the observed amplitude decay with distance, which makes the Q value computed from the Futterman operator questionable. This suggests that the Q value computed from the spectral‐ratio method is probably more reliable than values computed by other techniques.

Frequency dependence of Q in the mantle underlying the shield areas of Eurasia, Part 11: analyses of long period data

Summary. Long-period multiple S and ScS phases observed in northern Europe were analysed to determine mantle attenuation in the 0.02 to 0.2 Liz range under the Eurasian shield. Two groups of' events were used: deep FarEastern earthquakes and large earthquakes near the edges of the shield areas of Eurasia. The Q of the upper mantle under the Eurasian shield region was estimated in the time domain by taking amplitude ratios and in the frequency domain by taking spectral amplitude ratios among various arrivals. Under shield regions, SS/S amplitude ratios and multiple ScS amplitude ratios give ti - 2.5-3.0 s in the 0.02 to 0.20 Hz band. The results show that the upper mantle Q under this shield area is higher than the global average, but less than the Q values inferred from our studies of short-period data in the same area. Preliminary results suggest significantly higher /: under tectonic regions in the same frequency band.

A comparative ground response study near Los Angeles using recordings of Nevada nuclear tests and the 1971 San Fernando earthquake

AbstractA comparative ground response study at sites in the Los Angeles region is based on the extensive strong-motion data set recorded in the 1971 San Fernando earthquake and 159 three-component recordings of distant nuclear tests in Nevada. Amplitude spectral ratios computed for the nuclear test data over those frequency bands for which there is an adequate signal-to-noise ratio provide statistically stable estimates of the local ground response related to the type of local geologic conditions. Comparison of the strong-motion data recorded from the 1971 San Fernando earthquake with that recorded at 28 of the same sites for the nuclear tests show that the corresponding amplitude spectral ratios are statistically equivalent for most sites, provided reference stations are chosen to minimize effects on the earthquake data of azimuthal source variations and crustal propagation path. Statistical equivalence of the spectral ratios for the two types of data sources suggests that amplitude spectral ratios computed with respect to the appropriate reference station provide a first-order estimate of local ground response.

Study of daytime VLF spectral parameters of atmospherics as a function of azimuth

Abstract The spectral parameters ‘ group delaytime difference ’ (GDD) and ‘ spectral amplitude ratio ’ (SAR) have been determined from the phase and amplitude spectra of atmospherics received during the daytime from different azimuth angles during the period March 1976 to March 1977. The effect of anisotropy in propagation on the spectral parameters is that the GDD is greater in the west-east direction than in the reverse direction, and the SAR is lower for propagation in the west-east direction than in the reverse direction. These parameters depend on the direction of propagation of the atmospherics relative to the earth's magnetic field. The values obtained compare well with theoretical values reported elsewhere.

A method for the near-source anisotropy by the pair-event inversion of Rayleigh-wave radiation patterns

Summary. A novel method is proposed for retrieving the 3-D orientation of axes of symmetry of near-source anisotropy by a non-linear inversion of observed radiation patterns of seismic displacement spectra of Rayleigh waves. If faulting is generated within an anisotropic source region, body force equivalents for the faulting are in general not a double couple but the sum of three orthogonal dipole forces (Kosevich; Kawasaki & Tanimoto). As a result of the third dipole force, radiation patterns of Rayleigh waves are deformed, the deformation amounting to several per cent of those for an isotropic source medium. The non-linear inversion is carried out to find the optimum fault plane solutions giving the minimum square residual between observed and theoretical radiation patterns in some period range. In order to remove effects of heterogeneity along propagation paths, a pair-event scheme is involved in the inversion, which denotes taking spectral amplitude ratios and differential phases of the seismic displacement spectra of the pair-events having close hypocentres and different fault plane solutions. The uniqueness of the fault plane solutions of the non-linear inversion is afforded a proof by the Monte-Carlo experiment. The non-linear inversion is repeated for some possible types of symmetry of the near-source orthotropic anisotropy due to the preferred orientation of olivine crystals as mantle materials. Square residuals thus obtained are compared with each other to see which orientation gives the minimum. The method is applied to pair-events which occurred in the anomalous mantle beneath the Mid-Atlantic Ridge. This leads to a discovery that one type of symmetry of the preferred orientations with a-, b- and c-axes aligned vertical, parallel to and perpendicular to the trend (N11E) of the ridge axis, respectively, is most likely existing in the anomalous mantle.

Detection of erythrocyte membrane protein alterations in hereditary spherocytosis through the use of thermal stress: A spin label study

The membrane proteins of normal and hereditary spherocytosis have been labelled with a maleimide-analog nitroxide spin label and studied by electron paramagnetic resonance techniques. The spectral amplitude ratios from weakly and strongly immobilized labels differed slightly at 20° and 40°. Increasing the temperature to 47° and incubating for long time periods markedly accentuated the difference. It is suggested that the apparent differences in heat sensitivity between normal and hereditary spherocytosis erythrocyte membrane proteins reflect a latent structural alteration(s) of hereditary spherocytosis erythrocyte membrane proteins. Such structural alterations may result in altered functional behavior when the membrane is subjected to stress.

Enhanced sensitivity to slow motions using 15N-nitroxide spin labels

The rotational motion of an 15N nitroxide in diisobutylphthalate solutions at varied temperatures has been studied for correlation times from 10 −8 to > 10 −3 sec. Data from both conventional (first-harmonic, in-phase) and saturation-transfer (second-harmonic, out-of-phase) EPR are reported. For many spectral parameters, including linewidth, position of outer extrema, and ratios of spectral amplitudes in saturation transfer spectra, the 15N-nitroxide signals are sensitive to the same range of motions as are 14N-nitroxide spectra, but the 15N data offer the convenience of enhanced signal amplitude. In addition, the peak-to-peak separation of the low-field signal of 15N nitroxides in the conventional V 1 display has a smooth dependence on rates of motion from 10 7 to 10 4 sec and is measured in the most intense regions of the spectrum.

Ultrasonic Attenuation Measurement by Spectral Ratios Utilizing Signal Processing Techniques

A new approach using signal averaging and signal processing is described for measuring ultrasonic attenuation of compressional (P) and shear (S) waves in highly attenuative (low Q) materials. Broadband ultrasonic pulses in the frequency range of 0.7-1.1 MHz are transmitted through a specimen to be characterized for comparison to a reference with low dissipation. Attenuation is calculated from the ratio of spectral amplitudes which are corrected for diffraction effects. An experimental example is given based upon measurements made on dry polycrystalline sodium chloride (halite or rock salt) from Avery Island, LA. The measured QP of compressional waves is 46 ± 5.5 and QS of shear waves is 12 ± 0.3. This is a through transmission technique which gives results for materials so lossy that multiple echoes cannot be detected.

Lunar near‐surface shear wave velocities at the Apollo Landing Sites as inferred from spectral amplitude ratios

We reexamined the horizontal‐to‐vertical amplitude ratios of the long‐period seismograms to determine the shear wave velocity distributions at the Apollo 12, 14, 15, and 16 lunar landing sites. Average spectral ratios, computed from a number of impact signals, were compared with spectral ratios calculated for the fundamental mode Rayleigh waves in media consisting of homogeneous, isotropic, horizontal layers. The shear velocities of the best fitting models at the different sites resemble each other and differ from the average for all sites by not more than 20% except for the bottom layer at station 14. The shear velocities increase from 40 m/s at the surface to about 400 m/s at depths between 95 and 160 m at the various sites. Within this depth range the velocity‐depth functions are well represented by two piecewise linear segments, although the presence of first‐order discontinuities cannot be ruled out.

地表層付近におけるS波の減衰

Quality factor, QS, of attenuation for shear waves in the vicinity of ground surface is estimated by spectral analysis of body waves from micro-earthquakes observed at Mitsumata in Akita prefecture. The analysis is made for thirty earthquakes, in which S-P times are between 1 and 2sec.Based on some reasonable assumptions, it is found that the value of QS can be easily estimated from the average slope of spectral amplitude ratio between the direct P or S wave and the S or P wave converted at the boundary of the surface layer and the substratum. The result indicates that QS in the surface layer takes a value from 5 to 20 in the frequency range between 6 and 30Hz.

Lunar shear velocity structure at Apollo Sites 12, 14, and 15

Spectral amplitude ratios of horizontal-to-vertical motion produced on seismograms of meteoroid impacts at the Apollo 12, 14, and 15 sites, the Apollo 14 and 15 lunar-module impacts, and the Apollo 15 S4B impact show consistent differences among the recording sites. On the assumption that the motion represents predominantly fundamental-mode Reyleigh waves and that the compressional-wave velocity structure is similar to that derived in other investigations involving the Apollo seismic experiments, estimates are made of the shear-wave velocity structure under the three sites. Near-surface velocities are about 35 m/s at the three sites. The results for site 14 indicate an increase to about 100 m/s near the 8-m depth and to 200 m/s at the 38-m depth. Results for sites 12 and 15 show a smoother gradient and generally a greater velocity at a given depth than that indicated at site 14, reaching velocities of about 400 m/s near the 120-m depth. If the assumed P velocity structures are correct and if changes in S velocity coincide with changes in P velocity, then the ratio of these velocities decreases from 2.9 to 2.0 in the upper 19 meters at site 12, in the upper 38 meters at site 14, and in the upper 21 meters at site 15.

波形の相似な地震の発生と二, 三の問題

In the course of examining the wide dynamic range records of the aftershocks of the Tokachi-oki Earthquake of 1968, the seventy two shocks having similar wave forms out of about 50, 000 aftershocks were found by means of a simple but systematic data sorting. High correlations in P and later phases are clear not only among the events of the same order of magnitude but also among the events with considerably different magnitudes (M=1-4.6). The recurrent occurrence of such events can be traced for the length of about 50 days, which is much longer duration compared with so-called twin or triple shocks reported so far. The activity is, as a whole, a swarm type; however, it is composed of several foreshock-main shock-aftershock sequences. The determination of hypocenters for the several large shocks shows that the events are closely distributed in space within the range of 5km, which suggests that they have common spatial and dynamical characters. Hereby, the group of earthquakes having a similar wave character can be called an Earthquake Family, which may be the most basic unit of sequence of shocks.The nature of this family is characterized by a small b-value of 0.43 (total number=68) in the Gutenberg-Richter's formula: this result is in qualitative harmony with Mogi's explanation on b-value. The variations of P wave spectral amplitude ratio with magnitude are compared with the ω2- and ω3-theoretical source models and the ω2-model agrees well with the observed data in the frequency range between 4.6 and 16.2Hz. The average Qα in the crust is estimated to be 400 from the frequency dependence of the difference between the observed and theoretical amplitude ratio.

Eastern thunderstorms located by VLF atmospherics parameters

Daily variations of spectral atmospherics parameters, as measured by the atmospheric analyzer of the Heinrich Hertz Institut in Berlin‐Charlottenburg, are analyzed over an eight day period for long lasting storm centers within the octant NE‐E (receiving station Stockert near Bonn). The delay‐time difference and the amplitude ratio of spectral groups at 6 and 8 kHz were measured alternatively every 5 min. Within the first‐mode approximation these parameters are proportional to the propagation distance. Storm centers located by these parameter measurements are compared with synoptical observations. As theory indicates, the strong directional dependence of the GDD (group delay‐time difference) and the SAR (spectral amplitude ratio) on the propagation direction relative to the earth's magnetic field is verified by these measurements.

Causes of regional variation of magnitudes

abstract Data from the short and long-period seismographs at the NORSAR in Norway are used to investigate the discrimination of earthquakes and underground nuclear explosions using surface-wave versus body-wave magnitude (Ms versus mb). Earthquakes and explosions occurring within the western United States and recorded in Norway exhibit either anomalously large surface waves or anomalously low compressional body waves compared to events from central Asia. These data, as well as the results of other investigators, indicate an anomaly of 0.8 to 1.0 in Ms or 0.6 to 0.8 in mb or some linear combination of the two. The mechanism producing anomalously large Ms values from a region for explosions and the cause of lower mb values are investigated in terms of stress relaxation triggered by an explosion and regional variations in attenuation in the upper mantle beneath both the source region and the receiver region. The method of short-period amplitude spectral ratio is applied to the records of the waves from five deep events to determine the difference in attenuation beneath different receivers. The relative Q model inferred from these data for the upper mantle from 50 to 750 km depth is QLASA = 75, QTFSO = 175, and QNORSAR = 390. The circum-Pacific island arc exhibits an apparent source attenuation asymmetry. The data from the mid-Atlantic ridge indicate that strong attenuation may be associated with parts of the ridge. The relative difference of the Q model between LASA and NORSAR results in a difference in mb of 0.40 for distances of 60° to 80°, which agrees well with the observed differences in mb. We conclude that regional variations of attenuation in the upper mantle play an important role in regional differences in Ms versus mb relationship.

Seismic dissipation in deep seismic zones from the spectral ratio ofpP/P

The spectral amplitude ratio of the phases pP and P is influenced by source conditions and the crust-upper mantle structure near the source and at the recording station. The effect at the recording station is expected to be small because of nearly identical response of the recording instrument and its substructure to the phases that are incident below the crust at similar angles. The spectral ratios of pP/P for 48 earthquakes with focal depths ranging between 100 and 700 km are computed. Corrections for the effect of geometrical spreading and reflection in the outer shell above the focus are applied. The apparent mean Q values for the dilatational waves in the deep seismic zones of the upper mantle are computed from the slope of the adjusted spectral ratios. There is an indication of a minimum in the intrinsic Q near a depth at 200 km. The data of three South American earthquakes with focal depths between 120 and 200 km result in unacceptable negative values for (Q). Possible causes of these anomalous observations are discussed.