Search

Although the main problems with test ban treaty monitoring involve improving methods for discrimination of source types, there are certain problems of continuing interest in the field of yield estimation. In terms of short period body waves and the dependence of mb on yield, these involve the effects of anelastic attenuation, pP, nonlinear phenomena such as spall and influence of 3D structure on amplitude,. A related problem involves the transportability of yield scaling relation, hips between sites in different tectonic settings. We have carried out a number of studies directed at solving these problems. Much of the work has been presented in previous reports, but common themes among the studies are the use of a waveform analysis technique called intercorrelation and a focus on the use of the newly released data from internal Soviet stations. Tile first study reported here concerns the use of the intercorrelation approach on the onset of regional Pn waves to estimate yield. The yield dependence of the source time functions can be clearly observed and measured in the data. The importance of the method is that it provides a yield estimate based on an entirely new type of information (the waveform of the onset of Pn), and it is the type of information available from very small events at short distances. These are the types of data most crucial in the current treaty monitoring environment. The second study is an examination of broad-band yield scaling laws from all U.S. test sites and a correlation with them of results from Soviet sites.

The time and frequency domains are equivalent displays of seismic trace, information, though some qualities of the signal are more easily observed in one domain than the other. The relative frequency excitation of Lg, for instance, is most easily viewed in the frequency domain, but such waveform qualities as the sequence in which pulses arrive in the wave train or the sharpness of pulse onset are most easily studied in the time domain (Murphy and Bennett, 1982, Blandford, 1981). Because of the tremendous complexity of high frequency regional data, most attempts at using it for discrimination purposes have involved analysis of the frequency content of the various arrivals either through transforming selected windows or through multiple bandpass filtering. We report here on our initial attempts to explore the alternative and to discriminate events using those waveform characteristics most easily observed in the time domain. A second advantage of time domain analysis approaches is that they permit a deeper insight into the physical processes creating a seismic signal's character. For this reason, they can be more e3silv used to evaluate the transportabilty of a discriminant to varying geophysical and tectonic regimes. This is an especially important feature in the development of regional discriminants. The most prominent and successful spectral regional discriminants have been empirically developed. This means that they must be redeveloped and reverified in each new area. As we shall show in the following, through rigorous time domain analysis such features as regional depth phases can be identified and used to discriminate. Discriminants based on such simple physical features as source depth should be transportable anywhere. In work recently completed under the treaty verification program, we have proved that such time domain discriminants do exist. In analyzing a test discrimination data set from the western U. S., we have discovered that the onset of P_n is always v

The first part of this report presents a method for simultaneous inversion of near-field explosion waveforms for parameters of the source and structure. The inversion automatically provides information on the trade offs between parameters and error analyses of the resultant models. The source may be parameterized by any of several effective source functions. The structure is parameterized as a layered stack of gradients defined by the velocity at the top of the layer, the velocity gradient within the layer and the depth to the top of the layer. The gradient model is then discretized into homogeneous plane layers, with the layer thickness defined as the wavelength of the predominant period in the data, and synthetics are computed using asymptotic generalized ray theory. For each observed waveform, the residuals consist of the normalized cross-correlation coefficient between the data and synthetics, the difference in normalization factors and the time lag to the peak of cross-correlation. The second part of the report addresses the implications of choosing a particular representation of the RDP on the determination of teleseismic attenuation. This time-domain analysis is based on a near-field to far-field amplitude comparison. The near-field amplitude information is contained in the parameters of the explosion source model assuming amplitude information is contained in the parameters of the explosion source model assuming either the Mueller-Murphy or Helmberger-Hadley representations.

The first part compares the Mueller Murphy and Helmberger Hadley source representations and investigate whether there are any features of the near-field observations which tend to favor one over the other. The scaling relations in the two formalisms are such that the spectra are virtually indistinguishable between 2 and 5 Hz. Although the Mueller Murphy source is too rich in high frequency energy, a moderately low Q value for the near-field crust can make it agree with the observations. In the second part, the vertical particle velocity records for three nuclear explosions (DISCUS THROWER, MUDPACK and MERLIN) detonated in dry tuff of alluvium were modeled using generalized ray theory to determine the effective source functions. It was found that simple ray theory can be used to consistently predict the amplitude of the initial P wave for ranges greater than one depth of burial. At closer ranges the amplitudes are systematically overpredicted. Using the announced yields of the events, a scaling law was determined for source strength of nuclear explosions detonated in dry tuff or alluvium, and was used to predict the amplitudes from several other Yucca Flats events. The third part, introduces a method for the simultaneous inversion of near-field waveforms for source and structure parameters. The source may be parameterized by any of several effective source functions.

It is unlikely that regional data recorded at a newly installed array of stations could be used immediately to reliably discriminate events because of uncertainties about the local characteristics of wave propagation. However, the stations would immediately begin to record P sub nl data from small local events. This raises the possibility that the earthquake data base could be used to calibrate the array for discrimination purposes. The relationships is studied between long period P sub nl data and high frequency regional phases in the western U.S. This report examines the phase P sub g. The high frequency p sub g phase for small, shallow events typically appears as a complex arrival with a duration of several tens of seconds. This is much longer than the possible duration of the effective source time function even with depth phases taken into account. This phenomenon can be explained by synthesizing P sub g using generalized rays which is a practical computational procedure for a simple layer over a half space crustal model. If model responses are computed by progressively adding in rays with one reverberation in the crust, then two and so on, it is observed that a sequence of high frequency arrivals develops spanning the typical duration of P sub g. Each successive resonance phase is associated with a higher order reverberation in the crust. If each arrival contributes a typical amount of high frequency scattered energy, the long duration of P sub g is simple to understand. Keywords: Seismic discrimination.

Two regional discriminants are being developed and tested on western U.S. data bases. The first is based on the observation that the onset of the Pn waveform is stable in character and typically much different than most earthquake Pn waveforms. The results were reasonably consistent between stations. The average waveform at a fixed station is exceptionally consistent between the Pahute Mesa and Yucca test sites indicating that average Pn waveform is a stable function of time. Synthetics were computed both on the assumption that high frequency Pn is a turning ray in a smooth positive gradient in the lid and also that it is a true head wave. Effective pP is later than elastic time as is usually observed. This simple reflection and probably involves-spall. A set of earthquake Pn waveforms from the digital stations was collected. Earthquakes could be discriminated from explosions on the basis of the correlation of the Pn waveform with the average explosion Pn. The second discriminant is being developed around the Pg waveform. A suite of explosion Pg's was averaged to find which of its characteristics are most stable. Seismic Discrimination.

A regional event discriminant is being developed and tested based on the waveform of high frequency P sub n. The data base being used in the development consists of signals from explosions and earthquakes recorded on the western U.S. digital network. It has been discovered that at most stations the waveform of P sub n onset is remarkably stable and different from the corresponding waveforms from earthquakes. A forward modeling study of broad band explosion P sub n's revealed that the distinctive character of their waveform is caused by a strong p P sub n arrival. Depth phases from earthquakes arrive much later in the signal. It was found that a clear effective p P sub n arrival was present in all cases. However, it consistently arrives later than the predicted elastic time. For Pahute events, the amplitude of effective pP is close to the elastic predictions. For Yucca Valley, it is consistently larger indicating the effect of a site dependent nonlinear process in the source region. An appropriate value of t* for P sub n appears to be in the range of 0.1 to 0.2 seconds. The frequency content of the explosion P sub n energy indicates that it is caused by turning rays in the lid gradient rather than true head waves traveling on the crust mantle interface. The discrimination capacity of the P sub n waveform was measured quantitatively by correlating the average explosion P sub n trace with a data base of explosion and earthquake signals. Keywords: Synthetic seismograms, Nuclear explosion testing, Underground explosions.

Time domain waveform modeling studies of short period P sub nl and high frequency P sub n have been carried out in an attempt to improve event discrimination capabilities with regional data. The purpose was to develop discriminants with a well understood physical basis which could be reliably transported to different areas of the world. The study of P sub nl involved adapting methods developed to fit long period data for observations with periods as short as two seconds. To accomplish this, we have modified the wavenumber integration method for efficient computation of P sub nl. Velocity gradients like those typically observed near the free surface were of great importance. Gradients at lower depths such as at the crust-mantle transition are of much less significance. A simple layered model which accurately predicts regional P sub nl propagation from NTS to the surrounding digital stations were developed. It deterministically explains the waveshape of the P sub n and approximately the first 15 seconds of high amplitude energy reflecting past critical angle from the mantle. The modeling study of short period P sub n has led to the development of an effective and physically based regional discriminant; the underlying purpose of this project. It has been found that the waveshape of the P sub n is stable, similar at most stations and straightforward to model. (jhd)

Tenders are currently required to offload ULS (Unauthorized Long Supply) material at least every 90 days. This study evaluates variations in the timing of offloads, the value of the Economic Retention Level used in computing the ULS quantity, and various parameters used in computing an item's authorized inventory levels. Alternative offload policies were evaluated in terms of the impact on (1) dollar investment in on-hand plus due-in stock; (2) number of items offloaded; (3) dollar value of items offloaded; (4) number of resupply orders and Direct Turnover requisitions; (5) gross requisition effectiveness; (6) gross unit effectiveness; and (7) net total cost. Analyses were conducted for an FBM (Fleet Ballistic Missile) submarine tender and an attack submarine tender. The study identified seven alternative policies which reduced the number of current offloads by over 50% with no decrease in effectiveness and less than 2% growth in inventory dollar value. The most significant factor in these seven alternatives was an adjustment in the Economic Retention Level from the current value of $10 to $50 or $100. (Author)

This study evaluates alternative techniques for computing conventional submarine tender load lists, given specified performance goals. Areas that are evaluated include: (1) the use of alternative demand distributions; (2) the use of alternative techniques for controlling range; and (3) the use of alternative optimization models. Alternatives are evaluated separately for equipment-related and nonequipment-related items. The various techniques are evaluated using actual submarine demand data. The models are evaluated in terms of units effectiveness, requisitions effectiveness, and range effectiveness. A common model is recommended for computing conventional and FBM (Fleet Ballistic Missile) submarine tender load lists. (Author)

This study evaluates alternative techniques for computing FBM (Fleet Ballistic Missile) tender load lists, given specified performance goals. Areas that are evaluated include: (1) the use of alternative demand distributions; (2) the use of alternative techniques for controlling range; (3) the use of alternative optimization models; and (4) the use of minimum item protection level constraints. Alternatives are evaluated separately for equipment-related and non-equipment-related items. The various techniques are evaluated using actual FBM demand data. The models are evaluated in terms of units effectiveness, requisitions effectiveness and range effectiveness.

It is well established that near fields records of nuclear explosions can be analyzed to obtain detailed information about the seismic source function and its dependence on yield. This information is generally formulated in terms of parametrized models for the RDP and for the test site crustal structure. In this study, we review the results of forward modeling studies to obtain the source and structure parameters for Pahute Mesa. These models fit the observed near field records well, but there is some question as to how errors in crustal structure might affect seismic source parameters. Furthermore, there is a general need to be able to develop source-structure models in a consistent, unbiased fashion. To address these issues, we have developed a simultaneous inversion for source and structure parameters. In previous reports, we discussed the development of the method and applied it to Pahute Mesa data. In this report, we present an application of an improved technique for inverting for parameters in these types of problems known as the jumping method. A second problem with analysis of near-field records from explosions is that there is some question as to whether crustal materials respond in a linear anelastic or anelastic fashion or whether they have significant nonlinear response because of the high shear strain levels. Near-field data sets from several nuclear explosions and an earthquake are examined to address this question. It is shown that if nonlinear material response occurs it does not have a large enough effect to have a significant effect on data interpretation.

Laboratory evidence indicates that when strains exceed 10 to the -6 in media under low confining pressure, certain nonlinear processes begin to occur. They are manifested by a reduction in apparent Q. Since direct measurement of near field strain are rare, it has been difficult to evaluate whether or not this phenomenon is significant near to realistic seismic sources. A high frequency theory for computing stresses and strains using generalized ray theory has been developed. It can be shown using this theory that near field body wave strain pulses are closely related to near field velocity pulses. Large data bases of near field velocity records and crustal structures for modeling them are available. Transfer functions for transforming velocity pulses into estimated strain pulses can be computed by deconvolving theoretical versions of the former from the latter. In most cases they are delta-like functions. These transfers operators have been computed for a suite of velocity records from 5 Pahute Mesa Nuclear explosions ranging in yield from 155 to 1300 kt and for an Imperial Valley earthquake of M sub 0 = 0.6 times 10 to the 24 dyne-cm. For the explosion data base the strains were as high as 10 to the -3; 3 orders of magnitude higher than the level at which the laboratory data suggests that nonlinear effects become important. The earthquake data base indicated strains levels between 10 to the -6 and 10 to the -5. Because of the pressure dependence of the nonlinear phenomenon, it is probably only important in a thin layer near the surface of the earth.