ESEE  - Embayment Seismic Excitation Experiment: Collaborative Research with CERI and the U.S. Geological Survey

Table of  Contents
Significance of Project
Project Plan
      Velocity and Attenuation Structure of the Mississippi Embayment
    Excitation of High Frequency Surface Waves in the Embayment and the Q problem
    Implication of Low Velocity, High Q Embayment Sediments for Strong Ground Motions
    The Embayment Seismic Excitation Experiment - ESEE
    ESEE Scientific Targets

Back to ESEE Main Page


A 2600lb and a 5000lb explosion will be detonated within the New Madrid Seismic Zone (NMSZ) and Cooperative New Madrid Seismic Network in October to excite fundamental and higher mode surface waves within the sediments of the Mississippi Embayment.  The seismic network augmented by 8 temporary broadband stations and a three-component, short-period, small aperture seismic array will record these waves.  The low velocity sediments of the Mississippi Embayment have been long recognized as contributing to the seismic hazard due to shaking from large earthquakes in the NMSZ.  However, basic bulk properties of these sediments and their effects on seismic wave propagation are not well known.  In particular, there is reason to believe that attenuation of high frequency waves in embayment sediments has been overestimated by body wave spectral analysis methods.  Recently collected high frequency surface wave data from a small explosion show that attenuation is not as severe as predicted from previous body wave analyses.  Modeling of analog surface wave data from a large USGS explosion in 1991 likewise suggests that intrinsic Qp and Qs within the embayment sediments are significantly higher than previously believed.  The proposed explosion experiment will yield a unique surface wave data set that will illuminate velocity structure, attenuation structure, and wave propagation effects within the Mississippi embayment sediments allowing a greater understanding of the nature of strong ground motions from past and future New Madrid earthquakes.

Back to Table of Contents

Significance of Project

    This proposal is a request for second year funding for analysis of data collected from the Embayment Seismic Excitation Experiment (ESEE).  The experiment is planned for October 2002.  Currently, station and explosion siting is being performed as well as arrangements for explosives emplacement and detonation.  Second year funding is necessary for the working group to perform the data analysis to fulfill our scientific objectives.

    The unconsolidated sediments of the Mississippi embayment have long been conjectured as being important in evaluating the hazards from ground shaking induced by major earthquakes in the New Madrid Seismic Zone (NMSZ).  It is well known that sediments with low seismic velocities play a major role in amplifying seismic ground motions and that their non-linear, attenuative properties may also be important at high levels of strong ground motions.  Indeed, we show that the issue of seismic attenuation, parameterized by the seismic quality factors Qp and Qs, is largely unresolved in the NMSZ.  Although seismologists and engineers suspect that embayment sediments will tend to amplify ground motions, the trade-off between sediment impedance and attenuation may actually cause de-amplification of seismic waves in highly attenuating sediments.  It is imperative to determine both velocity and attenuation structure in the embayment to answer this basic question in hazards due to strong ground motions.
We also suggest that the sediments of the Mississippi embayment play a far more important role in creating seismic hazard through preferential excitation of high frequency surface waves for faults that rupture into the sedimentary layer.  Through theoretical considerations (see below), we expect that these high frequency surface waves will be 10 to 300 times larger in amplitude than body waves radiated from portions of the seismogenic fault below the sediments.  Thus, we suggest that the embayment sediments act as a significant wave guide which greatly increases the level of high frequency wave amplitude at local and regional distances.  This has profound implications in assessing the hazards from future earthquakes in the NMSZ and understanding the macroseismic effects and source processes of the 1811-1812 NMSZ earthquakes.

    We will be investigating velocity and attenuation structure of the embayment sediments by performing an active seismic experiment in the NMSZ this October (Figure 1 ).  Two large explosions will be detonated within the NMSZ and Cooperative New Madrid Seismic Network to excite fundamental and higher mode surface waves within the sediments of the Mississippi Embayment.  We emphasize that this is not a refraction experiment but an experiment for generating large Rayleigh and (possibly) Love waves.  Surface waves are very sensitive to shear wave velocity and their amplitude decay with distance yields first order estimates of intrinsic attenuation in the medium. Deployment and detonation of these sources is under the purview of Walter Mooney of the U.S. Geological Survey.  These explosions will be recorded by the CERI seismic network augmented by 8 temporary broadband stations and a three-component, short-period, small aperture seismic array, and an array of strong motion accelerometers deployed near the explosion sources.  This experiment will address directly the scientific issues of velocity structure, attenuation structure, and by inference, earthquake source excitation of high frequency surface waves.  The experiment will also yield basic seismological data on wave field heterogeneity caused by structural heterogeneity of the embayment.

This work addresses the following research priorities listed in the NEHRP Announcement for FY2003 for the Central/Eastern United States, Element II – Research on Earthquake Physics and Effects:

-    evaluate the effects of the Mississippi embayment structure (upper few km) on wave propagation (i.e., attenuation and the extent to which embayment structure acts as a wave-guide, reflects energy at its boundaries, etc.),
-    add constraints to attenuation relations using seismic network data from networks enhanced over the last several years and possibly from earthquakes in analog regions,
-    collect direct measurements of the physical properties of deep sediments of the Mississippi embayment and actual ground motions themselves.

Back to Table of Contents

Project Plan


Seismic hazard in the Mississippi embayment cannot be assessed accurately without a thorough understanding of seismic wave excitation and propagation.  This understanding must be based on well-constrained models and validated by direct observations ( Figure 1 ).  In addition to being required for predicting ground motions from future earthquakes, this knowledge underlies accurate interpretations of the geologic effects of strong shaking from historic and paleo-earthquakes, and thus also affects the accuracy of estimates of paleo-earthquake size and recurrence.   While most seismologists and geotechnical engineers would agree that the thick embayment sediment cover, exceeding 1 km in places, should significantly influence wave excitation, attenuation, resonances, and non-linear response, the precise nature of this influence remains almost pure speculation.  A paucity of appropriate data has hampered our learning, resulting from both a lack of instrumentation and low seismicity rates.  Although we can improve instrumentation, modifying the low seismicity rate remains beyond even the best seismologists’ capabilities!  This experiment proposes to do the next best thing – to compensate for the lack of earthquake sources by using artificial sources.

Active source experiments have been carried out in the Mississippi embayment before and, as described below, have significantly enhanced our understanding of the regional geological structure.  The novel element of the experiments we propose is that they use explosive sources and modern seismic recording technology to illuminate directly the structures that most strongly (and, at present enigmatically) influence the damaging wave fields of large earthquakes--the shallow sediments.  Our proposed experiment will differ significantly from past seismic experiments in that we will record signals over a broader frequency range that will allow us to analyze surface waves trapped in the sediments.  Past experiments relied on high frequency sensors without adequate dynamic range to record these surface waves with high fidelity.

The critical point is this: we have ample evidence (both observational and from modeling studies) that shallow explosions within the Mississippi embayment generate large surface waves.  Surface waves, because they are energy trapped for the most part within the sediments, contain a wealth of information about sedimentary properties that can be used to constrain models of the effects of those sediments on seismic wave propagation—and hence hazard.  We propose this work at a time when a well-distributed seismic network with adequate bandwidth exists for us to make use of this information.  In addition to the primary targets of sedimentary wave propagation, the experiments we propose should provide important data for secondary studies of crustal structure, non-linear seismic wave propagation, mode conversion at embayment boundaries, and others.

Back to Table of Contents

Velocity and Attenuation Structure of the Mississippi Embayment

The Mississippi embayment (Figure 1 ) is a trough plunging gently to the southwest  having developed since Mesozoic time (i.e. Stearns, 1957).  Since the Cretaceous, it has been accumulating a nearly continuous sequence of shallow marine, littoral, and estuarine sediments.  These sediments, no thicker than about 2 km in the south central portion of the embayment, remain poorly consolidated.  The most prominent seismic impedance contrast in the region is between the sediments, with shear velocities that average well below 1 km/s, and the hard Paleozoic floor with shear velocities that exceed 3 km/s. The depths to the base of the sediments has been mapped from a compilation of well-logs and reflection profiles (Dart and Swolfs, 1998), and industry reflection lines (i.e. Mihills and VanArsdale, 1999) providing a first-order 3-dimensional map of the sub-surface topography.
From the mid-1970s until the mid-1990s a permanent network of single-component, narrow band (short period), low dynamic range seismographs operated in the New Madrid seismic zone.  The temporary dense ‘PANDA’ network of 3-component short-period instruments operated for several years.  The addition of 3-components showed that the sharp impedance contrast at the base of the sediments leads to strong converted phases (Pujol et al., 1998).  The permanent network has only recently been replaced with a modern, 3-component dense network of both short-period and broad-band, higher dynamic range seismographs.

Already the broad-band data, albeit sparse, have confirmed the significant conversion of energy at the sediment/basement interface and have revealed strong site-resonances that clearly correlate with sediment thickness (Bodin and Horton, 1999).
Active source experiments in the region were designed to illuminate crustal-scale structures (Steinhart and Meyer, 1961; Mooney et al, 1983; Catchings, 1999).  One- and two-dimensional models of the shear- and compressional velocity structure of the embayment sediments have been derived from a variety of other data types.  These only minimally resolve lateral variations and in some cases are at odds with one another. Chen et al. (1996) interpreted the time differences between arrivals of converted Sp and S phases recorded by the PANDA network in terms of average sediment shear-velocities.  They inferred a range of 0.45 to 0.67 km/s, with increasing velocity correlating with increasing sediment thickness and presumably compaction.  These estimates contrast with those of Bodin and Horton (1999) who estimated an average shear velocity for the entire embayment of ~0.8 km/s, using observations of resonance peaks and assuming a quarter-wavelength relationship. Point-location measurements of compressional- and shear-wave velocity profiles from well-logs have been made in a few places, but only extending to 60 m in non-floodplain sediments and to 36 m at two floodplain sites (Liu et al., 1997).  These data indicate extremely high compressional/shear-wave velocity ratios, in excess of ~5, reflecting very low shear-wave velocities.

The regional scale, crustal compressional-wave velocity structure of the NMSZ below the sediments has been inferred using tomographic methods with earthquake P-wave arrival time data by Al-Shukri and Mitchell (1988).  They found that in regions overlying and containing active seismicity, velocities are lower by ~7% and ~14% respectively, relative to the rest of the NMSZ.  More recently, Vlahovic et al. (2000) employed PANDA network data to derive tomographic models of compressional-wave velocity and found similar correlations between low velocities and regions of active seismicity.  A number of long-distance refraction experiments have been conducted within and crossing the embayment. Catchings (1999) summarizes these and presents interpretations from the most recent experiment in which both shear-wave and compressional-wave data were collected over lines extending from Memphis to St. Louis.

Although critical for ground motion predictions, characterization of attenuation in the embayment is even more poorly constrained than the velocity structure.  A few studies have attempted to constrain compressional-wave attenuation, described by the quality factor Qp.  Al-Shukri et al. (1988) fit the attenuation parameter kp to P-wave spectra in the frequency band 10-25 Hz from earthquakes recorded by the earlier permanent seismic network.   These were interpreted in terms of whole-path Qp, which they inferred to be <100 for paths lengths shorter than r<25 km, 76<Qp<472 for 25>r>50 km, and 283<Qp<1088 for r>100 km.  Ray paths traversing regions of high seismicity had generally lower Qp estimates, but within the aforementioned ranges.  Al-Shukri and Mitchell (1990) used a similar dataset to estimate Qp in the sediments, and found it to be proportional to sediment thickness.  They also conducted a tomographic study to infer the lateral and depth variation of attenuation, dividing the region into 60x60 sq-km blocks with depths extending 0-5 km and 5-14 km.  They found 63<Qp<250 in the top layer and from 303<Qp<909 below.  In both layers the lowest values are found over or containing active seismicity, suggesting that high attenuation correlates with low seismic velocities.  Only one study of shear-wave velocity attenuation has been published to date. Liu et al.  (1994)  measured kp and ks from PANDA data in the band 5-25 Hz.  They assumed a two-layered model and found Qs=36 and Qp=59 for a homogeneous sedimentary layer and Qs=1020 and Qp=1199 below, noting that the latter particularly have very large uncertainties.  Kang and McMechen (1994) use relative amplitudes of arrivals in the refraction line of Mooney et al. (1983) to distinguish between intrinsic and scattering Q, obtaining estimates for Qp of~200 and Qs of 68.  Using a spectral ratio technique developed by Clouser and Langston (1991), Chen et al (1994) estimated sediment Qp and Qs from local Sp conversions seen in microearthquake data.  They suggested a range of Qp of 25-60 and Qs of 25-30 for the frequency band 2-25 Hz.

Back to Table of Contents

Excitation of High Frequency Surface Waves in the Embayment and the Q problem

In this section we show that high frequency surface waves are a natural consequence of shallow explosion sources.  Furthermore, we show that the surface waves yield first order information about sediment velocity and attenuation structure within the embayment that is difficult to obtain in any other way.  In fact, simple observation of surface waves from a large explosion shows that previous estimates of P and S wave attenuation determined for embayment sediments cannot be correct or, at least, cannot be applied to every location within the embayment.  Creating and recording surface waves through an explosion experiment offers a unique way to determine first order velocity and attenuation structure of the embayment sediments.  It also offers a direct look at the nature of wave propagation from future earthquake sources in the NMSZ.

The USGS detonated several large explosions in the Mississippi embayment in 1991 as part of a long-range refraction experiment (Catchings, 1999).  The design of the experiment required recording body waves on high-frequency sensors for obtaining accurate arrival time information so broad-band surface waves were not recorded.  However, CERI operated an experimental Guralp broadband borehole seismometer on its campus that recorded the last 5000lb shot of the series.  The shot (0022) was 24 km from the CERI campus near the Mississippi River ( Figure 1 ).
Although the recordings are analog, they nevertheless offer a fascinating view into the nature of wave propagation within the embayment waveguide.  The location and time of the detonation is accurately known and allows an analysis of average group velocities and amplitudes of the phases seen on the seismograms.  It proved impossible to accurately digitize these records because of the pen width of the analog recorder and high-frequency nature of the signals.  However, the amplitude envelopes were digitized to preserve the group arrivals and relative amplitudes.

Figure 2 shows the 3-component envelopes plotted together with arrows showing group arrivals with group velocity values(U=distance/arrival time).  The largest phase on these seismograms travels at ~2km/s.  The beginning of this phase is the P wave in the sediment waveguide seen previously in refraction experiments (e.g., Mooney et al 1983) and is denoted as the "acoustic mode" in Figures 3 and 4 .  There are major secondary arrivals, particularly on the vertical component, that arrive with velocities less than 0.6 km/s.  Because of their very low group velocity, these can only be multimode Rayleigh waves propagating in the sediments.  A careful look at the original broadband velocity records shows that these multimode Rayleigh waves have dominant periods of 1 - 2 seconds.  The "acoustic mode" is much higher frequency at 5-10 Hz.

A number of significant inferences about velocity and attenuation structure can be made by these very simple observations of group arrivals.  Synthetic seismograms were computed using a wavenumber integration algorithm (Barker, 1984) in an attempt to match the arrival time and relative amplitudes of these wave groups.  Figure 3 shows vertical component synthetics for four different velocity structures shown in Figure 5 .  Synthetics were computed using 4096 time points to a nyquist frequency of 10 Hz.  Vp/Vs ratios in the unconsolidated embayment sediments are known to quite high at 3 or greater (Dorman and Smalley, 1993; Liu et al 1991).  We use Nafe/Drake-type relations for Vp, Vs, and density suggested by Dorman and Smalley to construct sediment velocity models.

Starting with an ideal layer-over-halfspace model, we see that arrival times for the Acoustic and Shear modes are roughly correct but that the shape of the group arrivals are nothing like the data.  The acoustic modes build too slowly with time and the shear modes are dominated by a single arrival, which is actually the post-critical Shear wave reflection from the base of the sediments.  A two-layer model improves the duration of the shear modes but the acoustic modes still build too slowly with time.  Dorman and Smalley (1993) proposed an average model for embayment sediments from study of the Risco earthquake.  Their model has good attributes of three basic arrivals - acoustic, and 2 shear mode groups - but it does not reproduce the long-duration dispersion seen in the data.  Assuming that the dispersion is primarily controlled by the fundamental mode Rayleigh wave, we generated a perturbed model, the "Shelby" model, from group velocity partial derivatives.  While not perfect, this model does approximate the duration of the envelope of the data seismogram.  The acoustic group arrival is relatively impulsive and compact and the shear modes have 2 major group arrivals with appropriate durations.

This exercise proves that small changes in velocity structure have large effects on the wave field.  It is relatively difficult to match the duration and amplitude behavior of the shear modes.  (It is all the more difficult since the exact frequency content of these data is not known.)  A modal analysis shows that the configuration of fundamental and higher modes in the wave-guide is complex.  Even at relatively low frequency (~1 Hz) the fundamental and first higher modes interfere.  At higher frequency there is a skein of higher modes giving rise to the major acoustic and shear group arrivals.  Synthetic seismograms are required to understand the wave propagation.

It is interesting to examine the sensitivity of the synthetic seismograms to perturbations in the attenuation model.  Figure 4 shows a series of models in which Qp and Qs are varied in the sediments.  High Q’s (Qp = 500, Qs = 250) were assumed in trying the match the group arrivals.  In a study of Sp/S spectral ratios from microearthquakes in the embayment, Chen et al (1994) suggested much lower values (Qp=60 and Qs=30).  Q’s this low degrade the expected signal such that nearly all sediment modes are removed at 24km distance.  It is very difficult to reconcile these low Q’s with the simple observations of major, low group velocity arrivals in the data.  Having such low Q’s even in the upper most layer of the sediment model is not acceptable since most of the shear modes are attenuated after the 0.6 km/s arrival.  Modest levels of attenuation are possible, as seen in the lowermost synthetic of Figure 4 , but not at the levels suggested by previous attenuation studies.  The simple observation that these group arrivals exist in the GUR data suggests that Q in the embayment has been severely underestimated.  An experimental program in explosion source recording will be able to place meaningful constraints on both velocity and attenuation structure.

    Recently, we had the opportunity to record a small, buried test explosion (50lbs of high explosive) at a sand quarry near West Memphis, Arkansas.  An array of K-2 episensor FBA seismometers were deployed to a distance of 3 km from the shot.  Figure 6A shows ground velocity at these sites processed from the vertical acceleration data.  The data show large Rayleigh group arrivals out to 3 km with very low group velocities (0.5-0.15 km/s).  Synthetic seismograms that approximate the character of the data can be obtained through a simple modification of the Shelby model ( Figure 5 ) by placing a low velocity layer at the top ( Figure 6B ).  Figure 6C shows a model run where the sediments have relatively high Q’s and Figure 6D shows the case for low Q’s.  Because we are so close to the source, the Rayleigh waves are high frequency and sensitive to the upper structure of the embayment sediments.  Clearly the earth model must be perturbed to match the principal group arrivals better.  However, the low Q model again removes nearly all of the surface wave field from the synthetic records.

    These sparse surface wave data sets demonstrate several things.  First, explosions are efficient at setting up high frequency surface waves in the embayment that propagate to significant distances.  Secondly, they display very low group velocities indicative of propagation in low velocity, unconsolidated sediments.  The theoretical analysis of attenuation also shows that surface waves should not be observable if Qp and Qs are as low as 60 and 30, respectively.

We are looking into the attenuation problem as part of work on a present NEHRP grant.  Analysis of Sp conversions and direct S waves from small earthquake events in the embayment in conjunction with actual in-situ velocity depth functions suggests that the spectral ratio method (e.g., Clouser and Langston 1991) can be severely affected by site conditions in the upper 20m of the sediments.  Basically, Sp and S waves resonate in a low velocity zone near the surface.  P wave site resonance makes it appear that the S waves have been attenuated.  The net effect is an apparent low Qs and Qp.  At best, the Sp/S spectral ratio data constrain Q to the extent that Qs must be about three times the value of Qp.  The data do not yield any other information on Q in the sediments and it is not necessary to invoke frequency-dependent Q mechanisms to explain the disparity between body wave and surface wave attenuation.  In other words, there is no direct evidence for low Qs or Qp in embayment sediments at the present time from body wave spectral measurements.  There is certainly no quantitative reason to invoke complex, frequency-dependent Q models since so little is known of the actual wave propagation in the embayment.

    The surface wave data show that Q must be higher than previously believed.  This has important implications for the efficiency of wave propagation and the possible levels strong ground motions from earthquakes in the NMSZ.  The active source experiments performed this year will give first order information on velocity and attenuation structure important for understanding wave propagation in the embayment.

Back to Table of Contents

Implication of Low Velocity, High Q Embayment Sediments for Strong Ground Motions

There is ample geological evidence that fault rupture associated with paleo-earthquakes has offset sedimentary units of the embayment in a number of areas (e.g., VanArsdale et al 1992; Luzietti et al 1992; VanArsdale, 1999).  Indeed, it is very likely that thrusting during the 1811-1812 earthquakes caused faulting within the sediments and produced part of the Lake County uplift that dammed a tributary of the Mississippi creating Reelfoot Lake (Purser and VanArsdale, 1998).  Although geological evidence shows the existence of faulting in the sediments, recent microseismicity is confined to the depth range of ~5-12 km (Chiu et al 1992).  There have been few recorded events that lie in or near the base of the unconsolidated sediments.  Recent seismicity, therefore, may not be a good proxy for the nature of sources and surface wave excitation from large events that rupture into the sediments.  One notable exception is the observation made by Dorman and Smalley of long duration 1 Hz surface waves in the coda of the 1991 Risco earthquake.  This event was magnitude 4.6 but located at about 8 km depth.

Synthetic seismograms are calculated from a series of point sources in a plane-layered model to demonstrate the profound effects that the embayment sediment layer has on excitation and apparent amplification of seismic waves from sources that occur within it.  Figure 7 shows a suite of synthetic seismograms constructed for a thrust fault source (CLVD or compensated linear vector dipole source) at various source depths in the “Shelby” model.   Figures 7(A) and 7(B) show that, for constant source moment, these surface waves are 100 to 300 times larger than body wave amplitudes from a source at 2 km depth.  Theoretically, at least, the embayment sediments have the effect of boosting apparent source sizes by two orders of magnitude!

Distributed seismic sources, such as the events which occurred in 1811-1812, will have a more complex excitation behavior for these surface waves.  Assuming fault planes which are approximately 15 km wide (e.g., Chiu et al 1992) and for constant slip over the plane, excitation of strong, high-frequency surface waves will only occur for the upper 10% of the fault which lies in or near the sediment layer.  However, this part of the source will generate surface waves 100 to 300 times larger than body waves from the lower portion of the fault (0.1 x ~200 = 20 vs 0.9 x 1 = 0.9).  This simple calculation suggests that much of the hazard from strong ground motions in the NMSZ depends quite strongly on source depth and whether faults rupture the near-surface sediments. No doubt there are also details of source finiteness and rupture directivity that will affect surface wave excitation and phasing.  However, the source excitation effect appears to be a first-order characteristic of large sources in the NMSZ.

Back to Table of Contents

The Embayment Seismic Excitation Experiment - ESEE

    It is imperative that experimental data be collected to address the nature of surface wave propagation within the embayment sediments with considerable attention given to determining relevant parameters of basin geometry and physical properties (Vp, Vs, density, Qp, Qs) in three dimensions.  We will do this in October by detonating two moderate explosions in the NMSZ and recording the expected fundamental and higher mode surface waves on the existing CERI cooperative network with auxiliary broadband and short-period instrumentation.  Because of uncertainty in funding the first year of this project, wet Spring weather this year, scheduling delays at the IRIS/PASSCAL instrument pool, seasonal agriculture in the embayment, and coordination with the U.S. Geological Survey, the experiment is planned for October 2002 rather than the initially proposed March 2002. Technicians at CERI are currently permitting sites for the temporary station deployment and the explosions.  Walter Mooney of the U.S. Geological Survey is initiating work on site preparation, drilling, and explosion emplacement.  Now that a final time frame has been determined for the explosions, we will begin advertising for piggy-back experiments from other interested parties to take advantage of this unique seismic source.

    Figure 1 shows a schematic layout of the experiment with rough locations of shot points, 8 broadband PASSCAL stations, and a short-period 1km aperture seismic array imbedded within the CERI cooperative network.  We have taken advantage of natural profiles of CERI broadband stations to form one reversed profile and several unreversed profiles, one of which extends outside of the embayment.  Shot points will be in the midst of the permanent network and present day seismicity of the NMSZ.

The 8 broadband stations fill gaps between existing broadband network stations.  We propose to deploy the temporary broadband and array stations in the southern NMSZ near Memphis, TN, for the first 2600lb explosion.  This explosion is the smallest and can be emplaced in a single borehole.  We will start with the least expensive explosion to investigate the excitation and propagation of surface waves over relatively short distances.  Perhaps 2600lbs will be sufficiently large to produce surface waves that can be seen over the entire CERI network.  If so, then the second explosion can be reduced from our proposed 5000lb shot to only 2600lb.  This could save as much as $8K in emplacement and detonation cost for the larger explosion and could allow us to conduct a third explosion.

The second explosion, provisionally at 5000lb, will be detonated in the northern part of the NMSZ.  This is also the densest part the CERI network and we plan to run a profile to the northwest to investigate how high frequency surface waves interact with the embayment edge.  Five of the broadband stations in the south will be redeployed in the north along with the 11 element array.  Temporary stations will be obtained from the PASSCAL instrument pool at the time of the experiment.

Walter Mooney of the U.S. Geological Survey is in charge of contracting, siting, and logistics for the explosion work.  He has had extensive experience in the Mississippi embayment through past U.S.G.S. refraction experiments.

In addition to temporary deployment of broadband stations, we propose to install a 1km aperture seismic array for both the southern and northern explosions (Figure 1 ).   Figure 8 shows a design for this array consisting of 11 three-component S-13 short-period sensors.  The theoretical frequency-wave number domain array response is also shown.  This array is an essential part of the experiment in that it will be used to decompose  the seismic wave fields of body and surface waves sweeping across it.  Ten of the instrument sets will be obtained from the PASSCAL instrument pool. The eleventh set is available at CERI.

 The CERI cooperative network is the backbone of the experiment.  Consisting of over 60 short-period stations and 13 broadband stations within the NMSZ, this network will record data from the explosions that will be used for a variety of purposes.  The broadband stations will be used to study the nature of the surface wave propagation in conjunction with the temporary broadband stations.  We need an instrument passband between 0.25 and 2 Hz to record surface waves propagating in the embayment sediments.  CERI stations employ the CMG 40-T broadband sensor and we will be obtaining the same type of sensor from PASSCAL.  Although it is likely that the short-period stations will show surface waves, they are highly peaked at 4.5 Hz in an effort to record low magnitude seismicity which makes them generally unsuitable.

CERI will also deploy 8 K-2 strong motion accelerographs in a short (~1km) linear array near both explosions.  Data from these accelerographs will be used to investigate the near-field behavior of blast waves and analyzed to infer the explosion source function, near-field excitation of any high-frequency surface waves, short-offset deep reflections, and possible non-linear wave propagation effects in the high strain region near the source.

Back to Table of Contents

ESEE Scientific Targets
Our scientific priorities are listed in rank order as follows:

(1) Embayment Sediments Velocity Structure: Broadband observations of surface waves will be used to infer group velocity dispersion as a function of position throughout the recording area.  Coupled with phase velocity measurements made with the temporary seismic array and past P wave velocity measurements made through seismic refraction, it should be possible to constrain the shear wave velocity and Poisson’s ratio of the sedimentary column in the NMSZ.  We have available software for inverting dispersion and surface waveform data to obtain plane layered velocity models (Langston, 1995) and for creating a tomographic map of group velocity in the NMSZ to examine structure variations.

(2) Embayment Sediments Attenuation Structure: A detailed look at the spectral amplitude decay with distance of the multi-mode arrivals will be used to infer constraints on Qp and Qs in the sediments.  We will use a causal Q model associated with the wavenumber integration technique to explore the relationship between attenuation and velocity structure and to place hard constraints on seismic attenuation.  This will be possible because the observed surface waves are so slow they experience many wave cycles over each interstation distance, i.e., the attenuation effect is a first order effect for slowly propagating, high frequency surface waves in unconsolidated sediments.  Our experience with high frequency surface waves propagating in shallow structure has shown that the dispersion and waveform is a sensitive function of both Vp and Vs.  Thus, it is likely that Qp and Qs also control the attenuation behavior of the Rayleigh waveform.  It will be interesting to see if the experiment yields significant Love waves that can be used, in the absence of heterogeneity or anisotropy, to estimate Qs alone. We will investigate the possibility of inverting the attenuation measurements inferred along each profile to obtain a tomographic map of attenuation throughout the NMSZ.  We will investigate whether the higher mode arrivals constrain the attenuation structure through careful application of the full wave field synthetic seismogram to construct differential seismograms for attenuation parameters (Langston, 1995).  The determination of attenuation within the embayment has been an illusive goal because of the lack of appropriate data.  However, attenuation structure is a very important parameter in estimating the effects of strong ground motions from damaging earthquakes within the embayment.

(3) Basin Geometry and Regional Wave Propagation: The northernmost profile of broadband stations ( Figure 1 ) is designed to investigate how waves excited in the embayment sediments are stopped or continue propagating outside of the embayment structure.  If we are successful in recording the northern explosion inside and outside of the embayment, we feel that this dataset will be one of the more significant contributions to experimental seismology.  The issue of lateral heterogeneity is a basic one in regional wave propagation and it may be that the embayment sediments will cause a source amplification effect that may persist at all ranges.  This will have very important consequences in interpreting the co-seismic intensity effects of past large earthquakes in the NMSZ and estimating source size.  We will model these data using 2D and 3D wave propagation codes that incorporate the velocity and attenuation results obtained above.

(4)  Array Studies of Explosion Waves: Patterned after short-period small aperture arrays elsewhere (e.g., NORESS in Norway), we will be able to make a direct determination of phase velocity and azimuth of surface waves and determine structure under the array (e.g., see Vogfjord and Langston, 1996).  This will be indispensable in understanding the nature of the surface wave propagation in that estimates of phase velocity and group velocity will allow for an unambiguous interpretation of the dispersion.  For example, the group velocity of the fundamental mode is quite slow, 0.5 km/sec and less.  However, the theoretical dispersion curves show that phase velocities are quite high at ~2.2 km/s.  With such slow shear wave velocities in the sediments, this implies that the surface waves actually consist of near-vertically propagating shear wave reverberations in the sediment column.  This has important implications in understanding resonance effects in the sediment layer.  Analysis of body waves from the local seismicity will also yield important information on local wave scattering, converted phase wave propagation, and high resolution event locations.  We will also use the array to examine the coherence and character of the ambient noise that is important to basin resonance studies (Bodin and Horton, 1999).

(5)    Source Excitation Effects of the Embayment Sediments: The near field K-2 accelerograph array will be instrumental in understanding the source function of the explosions through waveform modeling of the initial P waveform.  We also envision the possibility, through detailed spectral and waveform analysis of the data, of investigating possible non-linear deformation effects of the near-surface sediments.  The explosions will probably be emplaced at ~50m depth in unconsolidated saturated sands which will liquefy near the source.  There may be non-elastic behavior of this material as the P wave and other phases propagate outward in the near field and are recorded by the K-2 array.  Array processing techniques can be used to determine dynamic strains and rotations (Gomberg et al 1999) and these will be compared to theoretical elastic wave propagation models.

Back to Table of Contents


Al-Shukri, H. and B. Mitchell, Reduced seismic velocities in the source zone of the New Madrid earthquakes, Bulletin of the Seismological Society of America, 78, 1491-1509, 1988.

Al-Shukri, H., B. Mitchell, and H.A.A. Ghalib, Attenuation of seismic waves in the New Madrid seismic zone, Seismological Research Letters, 59, 133-140, 1988.

Al-Shukri, H. and B. Mitchell, Three-dimensional attenuation structure in and around the New Madrid seismic zone, Bulletin of the Seismological Society of America, 80, 615-632, 1990.

Assimaki, Dominic, Frequency and pressure dependent dynamic soil properties for seismic analysis of deep sites. Ph.D. Thesis MIT, 2000.

Barker, J.S., A seismological analysis of the May 1980 Mammoth Lakes California earthquakes, Ph.D. Thesis, The Pennsylvania State University, University park, 278 pp., 1984.

Bodin, P. and S. Horton, Broadband microtremor observation of basin resonance in the Mississippi embayment, central U.S.:Implications for seismic hazard assessment  Geophysical Research Letters, 26, 903-906, 1999.

Casey, T., A. McGillvray, and P.W. Mayne, "Results of Seismic Piezocone Penetration Tests Performed in Memphis, Tennessee," Georgia Institute of Technology, GTRC Project E-20-E87, 1999.

Catchings, R.D., Regional Vp,Vs, Vp/Vs and Poisson's ratios across earthquake source zones from Memphis, TN to St. Louis, MO, Bulletin of the Seismological Society of America, 89, 1591-1605, 1999.

Chen, K., J.Chiu and Y. Yang, Qp-Qs relations in the sedimentary basin of the upper Mississippi embayment using converted phases, B.S.S.A., 84, 1861-1868, 1994.

Chen, K., J. Chiu, and Y. Yang, Shear-wave velocity of the sedimentary basin in the upper Mississippi Embayment using S-to-P converted waves, Bulletin of the Seismological Society of America, 86, 848-856, 1996.

Clouser, R. and C. Langston, Qp-Qs relations in a sedimentary basin using converted phases, B.S.S.A., 81, 733-750, 1991.

Dart, R.L. and H.S. Swolfs, Contour mapping of relic structures in the Precambrian basement of the Reelfoot rift, North American midcontinent, Tectonics, 17, 235-249, 1998.

Dorman, J. and R. Smalley, Low-frequency seismic surface waves in the upper Mississippi Embayment, Seismological Research Letters, 65, 137-148, 1994.

Ginzburg, A., W.D. Mooney, A.W. Walter, W.J. Lutter, and J.H. Healy, Deep structure of northern Mississippi Embayment, American Association of Petroleum Geologists Bulletin, 67, 2031-2046, 1983.

Gomberg, J., Pavlis, G., and P. Bodin, The strain in the array is mainly in the plane (waves below ~1 Hz), Bulletin of the Seismological Society of America, 89, 1428-1438, 1999.

Kang, I.B., and G. McMechan, Separation of intrinsic and scattering Q based on frequency-dependent amplitudes of transmitted waves.  Journal of Geophysical Research, 99, 23,875-23,885, 1994.

Liu, H., Y. Hu, J. Dorman, T. Chang, and J. Chiu, Upper Mississippi embayment shallow seismic velocities measured in situ, Engineering Geology, 46, 313-330, 1997.

Liu, Z., M.E. Wuenscher, R.B. Herrmann, Attenuation of body waves in the central New Madrid seismic zone, Bulletin of the Seismological Society of America, 84, 1112-1122, 1994.

Luzietti, E.A., Kanter, L.R., Schweig, E.S., Shedlock, K.M., and R.B. VanArsdale, Shallow deformation along the Crittenden County fault zone near the southeastern boundary of the Reelfoot rift, northeast Arkansas, Seismological Research Letters, 63, 263-275, 1992.

Mooney, W.D., M.C. Andrews, A. Ginzburg, D.A. Peters, R.M. Hamilton, Crustal structure of the northern Mississippi Embayment and a comparison with other continental rift zones, Tectonophysics, 94, 327-348, 1983.

Mihills, R.K. and R.B. Van Arsdale, Late Wisconsin to Holocene New Madrid seismic zone deformation. Bulletin of the Seismological Society of America, 89, 1019-1024, 1999.

Murphy, J.R., A.H. Davis, and N.L. Weaver, Amplifications of seismic body waves by low velocity surface layers, Bulletin of the Seismological Society of America, 61, 109-145, 1971.

Pujol, J., R. Herrmann, S. Chiu, and J. Chiu, Constrained joint location of New Madrid seismic zone earthquakes, Seismological Research Letters, 69, 56-68, 1998.

Purser, J.L., and R.B. Van Arsdale, Structure of the Lake County Uplift: New Madrid seismic zone. Bulletin of the Seismological Society of America, 88, 1204-1211, 1998.

Stearns, R.G., Cretaceous, Paleocene, and lower Eocene geologic history of the northern Mississippi embayment, Geological Society of America Bulletin, 68, 1077-1100, 1957.

Steinhart, J., and R.P. Meyer, Explosion studies of continental structure, Carnegie Inst. Wash Publ. 622, chapters 4, 5, and 7, 1961.

Tuttle, M.P., and E.S., Schweig, Towards a paleoearthquake chronology for the New Madrid seismic zone, U.S. Geol. Surv., NEHRP Ann. Rept., 17p., 1999.

VanArsdale, R., Displacement history and slip rate on the Reelfoot fault of the New Madrid seismic zone, Engineering Geology, 55, 219-226, 1999.

VanArsdale, R.B., Schweig, E.S., Kanter, L.R., Williams, R.A., Shedlock, K.M., and K.W. King, Preliminary shallow seisic reflection survey of Crowley’s ridge, northeast Arkansas, Seismological Research Letters, 63, 309-320, 1992.

Vlahovic, G.,  C. Powell, and J. Chiu, Three-dimensional P wave velocity structure in the New Madrid seismic zone, Journal of Geophysical Research, 105, 7999-8012, 2000.

Vogfjord, K.S., and C.A. Langston, Characteristics of short-period wave propagation in regions of Fennoscandia, with emphasis on Lg, Bulletin of the Seismological Society of America, 86, 1873-1895, 1996.

Wagner, G.S., Regional wave propagation in southern California and Nevada: observations from a three-component seismic array, Journal of Geophysical Research, 102, 8285-8311, 1997.

Back to Table of Contents