Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

ARTICLE IN PRESS

Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


journal homepage: www.elsevier.com/locate/soildyn

Shear wave velocities of Mississippi embayment soils from low frequency


surface wave measurements
Brent L. Rosenblad a,, Jonathan Bailey b, Ryan Csontos c, Roy Van Arsdale d
a
Department of Civil and Environmental Engineering, University of Missouri, Columbia, MO 65211, USA
b
U.S. Army Corps of Engineers, 1222 Spruce Street, St. Louis, MO 63103, USA
c
Ground Water Institute, University of Memphis, Memphis, TN 38152, USA
d
Department of Earth Sciences, University of Memphis, Memphis, TN 38152, USA

a r t i c l e in fo abstract

Article history: Deep unconsolidated sediments in the Mississippi embayment will influence ground motions from
Received 28 April 2009 earthquakes in the New Madrid seismic zone. Shear wave velocity profiles of these sediments are
Received in revised form important input parameters for modeling wave propagation and site response in this region. Low-
7 February 2010
frequency, active-source surface wave velocity measurements were performed to develop small-strain
Accepted 28 February 2010
shear wave velocity (VS) profiles at eleven deep soil sites in the Mississippi embayment, from north of
New Madrid, Missouri to Memphis, Tennessee. A servo-hydraulic, low-frequency source was used to
Keywords: excite surface wave energy to wavelengths of 600 m, resulting in VS profiles to depths of over 200 m. The
Surface waves average VS profile calculated from the eleven sites is in good agreement with common reference VS
Site response
profiles that have been used in seismic hazard studies of this region. The variability in VS profiles is
Shear wave velocity
shown to be associated with changes in formation depth and thickness from site-to-site. Using
New Madrid seismic zone
Dynamic soil properties lithologic information at each site, average formation velocities were developed and compared to
SASW previous studies. We found average VS values of about 193 m/s for alluvial deposits, 400 m/s for the
Upper Claiborne formations, and 685 m/s for the Memphis Sand formation.
& 2010 Elsevier Ltd. All rights reserved.

1. Introduction A common non-intrusive measurement method for determin-


ing VS profiles for geotechnical applications involves the analysis
The upper Mississippi embayment is a region of deep of actively generated surface waves [4,5]. Active-source surface
unconsolidated sediments that encompasses portions of Missouri, wave measurements have generally been applied to investigation
Illinois, Kentucky, Arkansas, and Tennessee. The depth of these depths of tens of meters due to a limited ability to generate
sediments (referred to as deep soil in this paper) ranges from low-frequency energy. In 2004, a unique field vibrator capable of
about 480 m near New Madrid, Missouri to nearly 1000 m near generating high forces at low-frequencies (o1 Hz) was commis-
Memphis, Tennessee [1]. These deep soil deposits overlie the New sioned as part of the National Science Foundation (NSF) Network
Madrid seismic zone and will affect the amplitude and frequen- for Earthquake Engineering Simulation (NEES) program. In May,
cies of earthquake ground motions in this region [2]. The New 2006 the NEES equipment was used at five deep soil sites in the
Madrid seismic zone is a zone of earthquake epicenters that Mississippi embayment as part of a study of active and passive
extend from near the town of Marked Tree, Arkansas to near surface wave measurement techniques for developing VS profiles
Charleston, Missouri (Fig. 1). One of the critical soil properties to depths of hundreds of meters. The VS profiles developed at
needed for seismic hazard studies of these deep deposits is the these five sites are presented and discussed in Rosenblad et al. [6].
small-strain shear wave velocity (VS) profile. Numerous VS In May 2007, surface wave measurements were performed using
measurements have been performed in the soils of the the NEES equipment at six additional sites located adjacent to
Mississippi embayment; however, these studies have generally seismic stations in the Mississippi embayment.
been confined to shallow depths (top 60 m) with relatively few In this paper, the variability of VS profiles in the Mississippi
studies investigating VS properties at depths below 60 m [3]. embayment to depths of 220 m is examined, and relationships
between soil formations and VS are developed. The average VS
profile calculated from the eleven VS profiles is compared to
common reference models used in seismic hazard studies of the
 Corresponding author. Mississippi embayment, and average soil formation velocities are
E-mail address: rosenbladb@missouri.edu (B.L. Rosenblad). developed and compared to values from previous studies.

0267-7261/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2010.02.010
ARTICLE IN PRESS
692 B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

Fig. 1. Map of study area showing the extent of the Mississippi embayment and the eleven measurements locations. Measurements were performed at Sites 1–5 in May
2006 and at Sites 6–11 in May 2007.

2. Overview of Mississippi embayment stratigraphy At locations in the upland region east of the Mississippi River
floodplain (Site 7 of this study, for example) the stratigraphy
The near-surface sediments of the upper Mississippi embay- consists of Pleistocene loess overlying Pliocene Upland Complex
ment consist of Holocene-age alluvial deposits in the lowland (Lafayette gravel) sand and gravel, which in turn overlies Eocene
(floodplain) regions, Pleistocene river terrace deposits locally strata [10], as shown in Fig. 2b. Lastly, at some locations (for
covered with loess, and Pleistocene-age loess deposits in the example Site 5 of this study) the near-surface soil consists of
upland regions [7]. The alluvial deposits of the Mississippi River Holocene-age alluvium from minor rivers overlying Eocene strata,
floodplain are composed of a near-surface silt/clay deposit, which as shown in Fig. 1c.
is underlain by sand and sand/gravel layers. The total thickness of The depth of the VS profiles determined in this study (220 m)
the alluvial deposits is typically in the range of 20–50 m. Beneath extended into but not below the Memphis Sand Formation at all
the near-surface sediments are Eocene-age silts and clays of the sites. Beneath the Memphis Sand are several hundred meters of
Jackson, Cockfield, and Cook Mountain Formations (also termed Paleocene and Upper Cretaceous deposits, which overlie Paleozoic
the Upper Claiborne unit). The combined thickness of the near- bedrock. A detailed presentation of the geology of the Mississippi
surface and Upper Claiborne deposits varies from 20 to 80 m embayment can be found in Van Arsdale and TenBrink [1].
beneath Memphis, Tennessee [8] near the southern extent of the
study area to over 180 m near New Madrid, Missouri [9] near the
northern extent of the study area. Beneath the Cook Mountain 3. Measurement sites
Formation is the thick Memphis Sand Formation, which is
composed of fine to coarse-grained sands with discontinuous Surface wave measurements were performed at eleven sites
clay lenses. The thickness of the Memphis Sand decreases from located in Arkansas, Tennessee, and Missouri, as shown in Fig. 1.
approximately 220 m near Memphis, Tennessee to 130 m near Field studies were performed at Sites 1 through 5 in May, 2006 [6]
New Madrid, Missouri. Nine of the eleven sites presented in this and at Sites 6 through 11 in May, 2007. Site locations were
paper are Mississippi River lowland sites. An example stratigra- distributed over a large extent of the northern embayment such
phy for Mississippi River lowland sites is presented in Fig. 2a. that different conditions of soil formation thickness and depth
ARTICLE IN PRESS
B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701 693

Fig. 2. Typical soil formations to depths of 250 m or greater for: (a) Mississippi River lowland sites, (b) upland site with near-surface Pleistocene-age loess, and (c) site in
Memphis with Wolf River alluvium at the surface.

were encountered. Most of the sites were located near existing interpolated values of formation tops of the silt/clay, sand, and
seismic stations operated by the Center for Earthquake Research sand/gravel layers in the Mississippi River alluvium; as well as the
and Information (CERI) at the University of Memphis (Sites 1, 3, estimated top of the Upper Claiborne unit and the top of the
6–11). Other considerations for site selection included accessi- Memphis Sand Formation. The unit tops within the alluvium and
bility of the large field equipment and the ability to obtain the the top of the Upper Claiborne unit were based on well
necessary permission/permits to perform the measurements. All information that was typically within 0.5–1.5 km of each site,
of the sites, except Site 7, were located on surface alluvial whereas the depth of the Memphis Sand was based on wells that
deposits. Table 1 presents the coordinates of the eleven site were typically within 4–9 km of the measurement sites.
locations and the estimated depth to Paleozoic bedrock at each
site. Bedrock depth estimates were determined from contouring
the top of the Paleozoic using petroleum exploration wells and 4. Measurement and analyses procedures
seismic reflection data in this region.
The lithology at each site was established from a detailed The work presented in this paper was derived from a larger
structural contour map of formation tops developed at the study of active and passive surface wave methods for deep VS
University of Memphis. The map was developed from geotechni- profiling. Therefore, at each site a variety of measurement
cal and geophysical well logs as well as seismic reflection lines techniques and data inversion methods were performed in-
performed in the embayment. This structural map provided cluding: the multi-channel f–k analyses method [11], the
ARTICLE IN PRESS
694 B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

Table 1
Site location information.

Site Site location Coordinates (deg.) Seismic station Depth to Paleozoic bedrock (m)

1 Mooring, TN 36.324N 89.566W MORT 703


2 Yarbro, AR 35.981N 89.915W – 783
3 Gosnell, AR 35.960N 90.016W GNAR 820
4 Lepanto, AR 35.614N 90.413W – 794
5 Memphis, TN 35.136N 89.846W – 840
6 Tennemo, MO 36.161N 89.579W TNMT 783
7 Glass, TN 36.269N 89.288W GLAT 766
8 Braggadocio, MO 36.205N 89.859W BRGM 714
9 Portageville, MO 36.450N 89.628W PENM 586
10 East Prairie, MO 36.717N 89.358W EPRM 451
11 Manila South, AR 35.784N 90.147W MSAR 847

Spectral-Analysis-of-Surface-Waves (SASW) method [4], the re-


fraction micrometer (ReMi) method [12], and the passive f–k
method using two-dimensional arrays [13]. The comparative
performance of these different methods is not the focus of this
paper, but is presented in Rosenblad and Li [14,15]. The VS profiles
presented in this paper were determined using the SASW analysis
procedure. The decision to use the SASW methodology is based
primarily on the implementation of an effective-velocity forward
model in the SASW inversion approach, as opposed to the typical
approach of using a fundamental-mode forward model in the
inversion. The effective-velocity approach requires no assumption
about the separation and identification of individual modes,
which was found to be advantageous for some sites where the
fundamental mode was not dominant over a portion of the
frequency range [16].

4.1. Field procedures and data processing

The SASW experimental procedure consists of developing an


‘‘experimental’’ dispersion curve from a series of two-channel Fig. 3. Experimental dispersion curve developed from surface wave measure-
phase measurements [4]. The wrapped phase measured between ments at Site 9 using seven receiver pairs.
receiver pairs is unwrapped by identifying 3601 jumps in the
phase plot and an ‘‘effective’’ phase velocity, VR, is calculated
using was assessed using the coherence function calculated between
the source signal and each receiver signal. Low-frequency ground
2pf
VR ¼ d ð1Þ motions produced by the NEES vibrator generally produced
f coherence values of 0.9–0.95 [6]. The ground vibrations were
where f is frequency, f is the unwrapped phase angle, and d is the measured using Mark Product L4 geophones and recorded
spacing between receivers. Data from a single receiver pair with using a VXI Technology dynamic signal analyzer (Model 1432A)
spacing, d, can be used to determine surface wave velocities over a for the 2006 study and a Data Physics dynamic signal analyzer
range of wavelengths from about d/3 to 2d. Therefore, to develop a (Mobilyzer II) for the 2007 study. Source and geophone locations
dispersion curve over a broad range of wavelengths, several were surveyed in using a total station and the geophones were
receiver pairs with different spacing are needed. For this study, buried to a depth of about 15 cm to minimize wind noise. An
effective-velocity dispersion curves were generated using 7 to 10 example experimental dispersion curve determined from SASW
receiver pairs with spacing that ranged from as low as 3 m to analyses of seven receiver pairs at Site 9 is shown in Fig. 3. The
about 300 m, resulting in dispersion curves with wavelengths manual phase unwrapping interpretation used in SASW analysis
from 1 m to about 600 m. has been shown to be ambiguous for some profile conditions [18].
High-frequency surface wave energy recorded with receiver Therefore, it should be noted that the experimental dispersion
pairs spaced at less than 20 m was excited using an instrumented curves developed and used in this study were verified to be
sledge hammer source. Low-frequency energy recorded using consistent with those developed using multi-channel wavefield
receiver pairs with large spacing ( 420 m) was excited using the transformation methods [16].
NEES low-frequency vibrator (‘‘Liquidator’’), which is capable of
exciting high force levels at frequencies down to less than 1 Hz
[17]. The NEES vibrator was operated in a stepped-sine mode 4.2. Data analysis
where the source was driven with a sinusoidal excitation that was
stepped from about 12 Hz down to 0.8 Hz. (Some measurements In SASW analyses, VS profiles are generated by fitting a
were attempted at frequencies below 0.8 Hz, but the output was theoretical dispersion curve to the experimental surface wave
too low to generate reliable dispersion data.) Approximately 50 to dispersion curve using an iterative forward modeling or inversion
200 cycles were applied at each frequency before stepping to the procedure. For this study, VS profiles were generated for each site
next frequency. The quality of the measurement at each frequency using the surface wave inversion program WinSASW2, developed
ARTICLE IN PRESS
B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701 695

at the University of Texas at Austin [19]. As noted above, the


approach used in this program is to calculate an effective-velocity
dispersion curve as opposed to the more conventional approach of
using single or multiple modal velocity dispersion curves. The
distinction between modal and effective velocity models is
discussed in detail in Roesset et al. [20]. The forward model used
in WinSASW2 calculates the surface displacement at each of the
receiver locations due to a harmonic surface disk load applied at
the source location. For each frequency, the effective phase
difference between two receiver locations due to the contribu-
tions of body waves and multiple surface wave modes is
calculated. This approach essentially simulates the field SASW
measurement and produces a separate theoretical dispersion
curve for each receiver pair.
The inversion program requires a starting model consisting of
layer thickness, VS, compression wave velocity (or Poisson’s ratio),
mass density, and material damping. All of these parameters,
except for VS, are held constant during the inversion. Compression
wave velocities of 1600–1800 m/s were assumed for soils below
the water table (Poisson’s ratio of 0.25 was assumed for soils
above water table), mass density was assumed to be 1900 kg/m3,
and material damping was assumed to be 2%. These values are
consistent with published values for the soils in the Mississippi
embayment [21–23]. The inversion procedure was first applied
using a general profile layering with layer thickness that
increased from less than 1 m near the surface to 45 m at depths
beyond 200 m. This inversion was performed without a priori
knowledge of formation thickness and depth at these sites. A
second inversion was performed at each site using thicker layers
and boundaries consistent with the formation depths estimated at
each site, as discussed later. The starting VS profile for each
analysis was generated using the procedure described in Joh [19].
This approach uses a preliminary inversion step to develop
several possible starting models based on analysis of the phase
velocities in the experimental dispersion curve. The preliminary
model with the lowest root-mean-square (RMS) error is used as
the starting model in the inversion procedure.
The inversion procedure used in WinSASW2 is a maximum
likelihood approach [24,25] which requires estimates for model
and data covariance. For this analysis standard deviations of 5% of
the model and data values were used. The program solves for each
of the individual dispersion curves (for each receiver pair) using
the starting model parameters, and calculates the RMS error
between the experimental and theoretical curves. The model
parameters are updated using the sensitivity matrix from the
forward equation and the maximum likelihood cost function. The
procedure is repeated until an acceptable fit is achieved, as
indicated by a low RMS value and negligible changes in the fit
with additional iterations. An example of starting and final VS
profiles and the respective dispersion curves are shown in Fig. 4.
The inversion analysis was performed using an initial profile
depth of 400 m (2/3 of the maximum wavelength). A manual
depth resolution analysis was performed by perturbing the lowest
layer by 720% and observing the change in the dispersion fit at
the longest wavelength. If a velocity change of less than 5% was
observed at the longest wavelength, the lowest layer was
removed and the halfspace velocity was assigned to the next
highest layer in the profile. This procedure was repeated until the
theoretical dispersion curve was shown to be sensitive to changes
in the lowest layer. In most cases this depth was in the range of
200–250 m. To indicate the sensitivity of the dispersion fit to
changes in the individual layers, a manual procedure was Fig. 4. Data analyses from Site 9 showing (a) experimental dispersion curve and
followed where the velocity of each layer was increased and theoretical dispersion curve from starting model, (b) final match between
decreased in 5% increments until a 10% change in RMS error was experimental and theoretical dispersion curves, and (c) starting and final VS
observed. The range in layer velocities is indicated with bars in profiles.

the profiles presented in Fig. 5.


ARTICLE IN PRESS
696 B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

Fig. 5. Shear wave velocity profiles developed at Sites 1–11 (a–k, respectively) and estimated lithology at each site. Bars on profiles indicate range in individual layer
velocities to cause 10% change in RMS fit to the dispersion curve (dashed lines indicate 420% change in layer velocity).

5. Results and discussion

5.1. Shear wave velocity profiles

Shear wave velocity profiles developed from the surface wave


measurements are presented along with the estimated lithology
at each site in Figs. 5a–k. The VS profiles shown in Figs. 5a–e (from
the 2006 study) are the same as those reported in Rosenblad et al.
[6], while Figs. 5f–k show the profiles developed from the 2007
study. The VS profiles developed at these eleven sites generally
show three distinct regions of VS values consisting of: (1) a high-
gradient, low-velocity near-surface region, (2) an intermediate
region with shear wave velocities of about 400–450 m/s, and (3) a
high-velocity region at depth with velocities increasing to 600–
800 m/s.
Fig. 6. Average VS in top 30 m (VS30) at Sites 1 through 11.
The average shear wave velocities measured in the top 30 m
were very similar at each of the sites. Fig. 6 presents a plot of the
average VS in the top 30 m calculated using
where di is the thickness of layer i, V Si is the VS of layer i, and V S30
P
n
di is the average VS in the top 30 m. As seen in Fig. 6, the VS30 values
V S30 ¼ 1
ð2Þ ranged from 180 to 224 m/s at the alluvial sites, while the upland
P
n
di site with older near-surface loess deposits (Site 7) had a slightly
Vsi
1 higher value of 247 m/s. With the exception of Site 6, the sites are
ARTICLE IN PRESS
B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701 697

estimating the velocity profile using a single power function with


depth (or effective stress) would predict the same depth where VS
exceeds 500 m/s for each site. These results, therefore, demon-
strate the importance of lithology on the local VS profile.

5.2. Comparison with reference VS profiles for the Embayment

Seismic hazard studies in the Mississippi embayment require


knowledge of the small-strain VS profile. A reference VS profile for
the embayment was developed by Romero and Rix [3] from a
compilation of existing VS profiles and has been used in seismic
hazard studies of the embayment [28,29]. Using the eleven VS
profiles developed from this work, the arithmetic mean and
standard deviation were calculated as a function of depth and
compared to the Romero and Rix reference profile, as shown in
Fig. 8. Also, shown in Fig. 8 is a VS profile presented by Dorman
and Smalley [30] that was developed from p-wave well velocities.
Dorman and Smalley [30] analyzed surface waves propagating
between Risco, Missouri and Memphis, Tennessee (similar extent
as our study area) and found the data to be consistent with the
profile shown in Fig. 8. As can be observed in Fig. 8, the average
profile developed from this study is in generally good agreement
with the profiles developed from these previous studies.
An association between soil formation and VS values can be
inferred from the fluctuating coefficient of variation (COV) values
with depth, shown in Fig. 8. At sites without abrupt changes in
velocity with depth, it is expected that COV values will be highest
near the surface (where variable moisture content conditions
dominate) and trend to lower values at greater depths as the
influence of effective stress dominates. In this study, the COV
value was highest at the surface (as expected) but fluctuated
between about 0.1 and 0.2 at greater depths. From examining
Figs. 5 and 8, it can be observed that depth intervals with low COV
values are consistent with depth ranges where the formation type
was generally the same at each site. For example, in the depth
Fig. 7. Comparison of shallow (top 30 m) VS profiles determined at Sites 4 and 9
with VS profiles determined from nearby downhole measurements in Marked Tree,
range of 50–100 m, where the COV values are low (0.11–0.14) all
AR and Risco, MO [27]. of the sites (except Site 5) have common soil formations (Upper
Claiborne unit). The COV values increase below this depth, as

classified as site class D according to the IBC code [26]. The results
from this study agreed well with shallow downhole
measurements performed by Liu et al. [27] near Sites 4 and 9
(Fig. 1) as shown in Fig. 7.
Below the soft, near-surface deposits, the VS profiles generally
show a distinct transition to a fairly uniform VS value of about
400 m/s (Fig. 5). When compared with the estimated soil profiles
at each site, it is apparent that this velocity zone is associated
with the Upper Claiborne unit (Jackson, Cockfield and Cook
Mountain Formations). For example, at Sites 1, 6, and 9, where the
Upper Claiborne unit extends to depths of about 150 m, the VS
profile flattens out and remains below about 500 m/s, while at
sites where these deposits are much thinner (3, 4, 5, and 11) the
velocities at depths of 100–150 m are in the range of 600–800 m/s.
The only clear exception to this observation is Site 10 where the
intermediate low-velocity region appears to extend well below
the expected base of the Upper Claiborne unit.
The velocity transition where values begin to increase from
400 to 450 m/s up to values of 600–800 m/s is consistent with the
expected depth of the top of the Memphis Sand. With the
exceptions of Sites 8 and 10, the top of the Memphis Sand
formation is generally consistent with the depth ( 710 to 15%)
where VS first exceeds 500 m/s. For example, at Site 5 the depth to
the top of the Memphis Sand and the depth where VS first exceeds Fig. 8. (a) Comparison between average VS profile calculated from eleven
measured profiles in this study, reference VS profile from Romero and Rix [3],
500 m/s is about 50 m, while at Site 9, VS does not exceed 500 m/s
and profile from Dorman and Smalley [30]. Bars indicate 7 one standard
until a depth of about 150 m, which is consistent with the much deviation. (b) Coefficient of variation versus depth calculated from the standard
deeper Memphis Sand deposit at this site. A typical approach of deviation and arithmetic mean of the eleven profiles.
ARTICLE IN PRESS
698 B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

some of the sites move into the Memphis Sand formation, and tended to group in separate velocity ranges with low COV values
then ultimately decrease again below 150 m, where all of the sites (with the exception of the gravel unit). In the case of the Upper
are in the Memphis Sand Formation. The relationship between Claiborne unit, exclusion of the much shallower velocity value
soil formation and VS is investigated further in the following from Site 5 brings the COV value down to 0.11. It should also be
section. mentioned that the greater VS variability in the gravel is possibly
due to localized iron oxide cementation in this layer like that
which we have observed in gravel quarries in the Upland
5.3. Soil formation velocities Complex.
Gomberg et al. [8] performed an extensive study of formation
The inversion analysis presented above was performed using a velocities from shallow VS measurements around Memphis, TN.
generic profile layering for each site and no a priori knowledge of One of the conclusions they reached from that study was that the
the subsurface layering. To better estimate average formation lithologic model was well correlated with VS and could be used as
shear wave velocities, the inversion procedure was repeated at a proxy for VS values. Also Williams et al. [31] developed
each site using thicker layers with boundaries at the estimated formation velocity estimates from refraction and reflection
formation tops for the Upper Claiborne and Memphis Sand. The measurements around the embayment. A comparison of the
layering in the alluvial deposits remained unchanged, but the average VS values developed from this work to those developed by
multiple layers in the Upper Claiborne and the Memphis Sand Gomberg et al. and Williams et al. is presented in Table 2.
were replaced by one or two layers. Fig. 9 shows an example from Comparisons are shown for the alluvium (loess layer at Site 7
Site 9 of the comparison between the original inverted VS profile was excluded from this average), the Upper Claiborne unit
and the revised VS profile using the formation boundaries and (Jackson/Cockfield/Cook Mountain Formation), and the Memphis
thicker layers. The only site where it was not possible to obtain an Sand. (The Lafayette gravel, also termed Upland Complex,
equivalent fit to the dispersion curve (in terms of RMS) with the around Memphis that is reported in the Gomberg et al.
estimated formation boundaries was at Site 10. In this case, we and Williams et al. studies is only present at one of our sites,
used a greater thickness for the Upper Claiborne unit (to a depth therefore an average velocity comparison is not presented for this
of 140 m) than was indicated for this site. Local and relatively material.)
minor erosion and deposition are common in these terrestrial The average value determined for the alluvium in this study
deposits and may not be detected in interpolated contour surfaces (193 m/s) fell between the values reported by Gomberg et al.
[8]. Therefore, it is possible that the differences observed at Site (171 m/s) and Williams et al. (206 m/s). The average value we
10 could be attributable to scour of the Upper Claiborne into the report for the Upper Claiborne unit (399 m/s) is also in good
underlying Memphis Sand or clay deposits at the top of the agreement with previous studies (about 3% lower than Gomberg
Memphis Sand. et al. and 12% lower than Williams et al.). The average velocity we
Using the revised profiles, formation velocities were deter- estimate for the Memphis Sand (685 m/s), however, is signifi-
mined at each site for the silt/clay and sand alluvium layers, the cantly higher than values reported by Gomberg et al. (530 m/s) or
sand/gravel layer, the Upper Claiborne unit, and the Memphis Willliams et al. (587 m/s). It should be noted that values from both
Sand Formation. For formations with multiple profile layers (such the Gomberg et al. and Williams et al. studies are based primarily
as the alluvium) the average velocity for the layer was calculated on shallow measurements performed in and around Memphis and
using Eq. (2). Fig. 10 presents the average VS values versus depth are expected to yield lower velocities due to the shallower depth.
and VS histograms for each of the four geologic units. The units The higher Memphis Sand velocities from this study are not
inconsistent with other measurements performed in the embay-
ment. For example, Fig. 11 presents suspension log data from a
deep Memphis Light Gas and Water (MLGW) well located in
Memphis, Tennessee (data provided by Glenn Rix, Georgia
Institute of Technology). The suspension log data show
variability in the VS values that is not observed in the surface
wave profiles due to the greater vertical resolution of the localized
suspension log measurements. Importantly, these data show the
top of the Memphis Sand to be consistent with an abrupt increase
in VS from about 350 m/s in the Upper Claiborne to near 600 m/s
in the Memphis Sand. Over the depth range of 77–220 m, the
average VS in the Memphis Sand from the suspension log data is
632 m/s (calculated using Eq. (2)), which falls within the range of
values determined in this study. Secondly, Street et al. [32]
present results from shear wave refraction and reflection
measurements performed at 10 sites in Arkansas distributed
from east to west across the Mississippi embayment. At each site
the VS profile showed a high-velocity layer at depths between 50
and 100 m with VS values ranging from 601 to 793 m/s. The
average velocity of this layer from these 10 sites is 690 m/s with a
standard deviation of 55 m/s, which is consistent with the high
velocity values reported in this study. In addition, Street and
Woolery [33] performed measurements in Southeast Missouri,
which also showed shear wave velocities at depths below 100 m
Fig. 9. Example comparison for Site 9 between VS profile determined using the of 600–800 m/s at most sites. One of these locations was within
generic layering (with no a priori knowledge of formation boundaries) and VS
profile determined using layering consistent with the depth to the top of the
about 800 m of the measurements we performed at Site 10. As
Upper Claiborne unit and Memphis Sand. Bars indicate change in layer velocity to shown in Fig. 12, these two methods produced very similar results
cause a 10% change in RMS fit to the dispersion curve. with velocities at depth of over 700 m/s.
ARTICLE IN PRESS
B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701 699

Fig. 10. Average formation velocities and histograms determined for (a) silt/clay and sand, (b) sand/gravel, (c) Upper Claiborne (Jackson/Cockfield/Cook Mountain
Formations), and (d) Memphis Sand Formation.

Table 2
Comparison of average formation velocities (standard deviation is shown in
parenthesis).

Study Average formation velocity (m/s)

Alluvium Upper Claiborne Memphis sand

Rosenblad et al. 193 (14) 399 (62) 685 (83)


Gomberg et al. [8] 171 (24) 413 (105) 530 (134)
Williams et al. [31] 206 455a 587

a
For Jackson formation/Lafayette gravel.

6. Conclusions

This study has provided VS profiles to depths of over 220 m at


eleven sites distributed around the northern Mississippi embay-
ment. The average profile generated from these eleven sites is
shown to be in generally good agreement with common reference
profiles for this region. The coefficient of variation of the average
profile was observed to fluctuate with depth in a manner that is
consistent with changes in soil formation depths from site-to-site.
Changes in VS profiles with depth at these sites are shown to be
consistent with the expected depth to formation tops, specifically
the top of the Upper Claiborne unit and the top of the Memphis
Sand. These results support the assertion made by Gomberg et al.
[8] that shear wave velocities are strongly correlated to lithology
in the embayment. Average shear wave velocities for the alluvium
and Upper Claiborne unit (Jackson/Cockfield/Cook Mountain
Formations) from this study are consistent with the results of
Fig. 11. Profile of VS from suspension log at Memphis Light Gas and Water
previous studies of shear wave velocities of these formations
(MLGW) well in Memphis, TN, showing measured formation velocities in the
[8,31]. However, the results from this study indicate that the Jackson/Cockfield/Cook Mtn (Upper Claiborne) and the Memphis Sand formations
average VS in the Memphis Sand ( 685 m/s) is higher than (data provided by Glenn Rix, Georgia Institute of Technology).
ARTICLE IN PRESS
700 B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701

[4] Stokoe II KH, Wright S, Bay J, Roesset JM. Characterization of geotechnical


sites by SASW method. Geophysical characterization of sites. Oxford: Oxford
& IBH Pub. Co.; 1994. p. 15–25.
[5] Park CB, Miller RD, Xia J. Multichannel analysis of surface waves. Geophysics
1999;64(3):800–8.
[6] Rosenblad BL, Li J, Menq F-Y, Stokoe KH. Deep shear wave velocity profiles
from surface wave measurements in the Mississippi embayment. Earthquake
Spectra 2007;23(4):791–808.
[7] Saucier RT, Snead JI. Quanternary geology of the Lower Mississippi Valley:
Louisiana Geological Survey; 1989, Map Scale 1:1,000,000.
[8] Gomberg J, Waldron B, Schweig E, Hwang H, Webbers A, Van Arsdale R, et al.
Lithology and shear-wave velocity in Memphis, Tennessee. Bulletin of the
Seismological Society of America 2003;93(3):986–97.
[9] Crone AJ. Sample description and stratigraphic correlation of the New Madrid
Test Well New Madrid County. Report No.: 81–426, US Geological Survey,
1981.
[10] VanArsdale RB, Bresnahan RP, McCallister NS, Waldron B. The Upland
Complex Of the central Mississippi River valley: its origin, denudation,
and possible role in reactivation of the New Madrid seismic zone. In:
Stein S, Mazzotti S, editors. Continental intraplate earthquakes: science,
hazard, and policy issues. Geological Society of America Special Paper 425,
2007. p. 177–92.
[11] Rix G, Hebeler G, Orozco MC. Near-surface VS profiling in the New Madrid
seismic zone using surface-wave methods. Seismological Research Letters
2002;73(3):380–92.
[12] Louie JN. Faster, better: shear-wave velocity to 100 meters depth from
refraction microtremor arrays. Bulletin of the Seismological Society of
America 2001;91(2):347–64.
[13] Zywicki DJ. Advanced signal processing methods applied to engineering
analysis of seismic surface waves: PhD thesis, Georgia Institute of
Technology, 1999.
[14] Rosenblad B, Li J. Comparative study of Refraction Microtremor (ReMi) and
active source methods for developing low-frequency surface wave dispersion
curves. Journal of Environmental and Engineering Geophysics 2009;
13(3):101–13.
[15] Rosenblad BL, Li J. Performance of active and passive methods
for measuring low-frequency surface wave dispersion curves. Journal
of Geotechnical and Geoenvironmental Engineering 2009;135(10):
1419–1428.
[16] Li J. Study of surface wave methods for deep shear wave velocity profiling
applied in the upper Mississippi embayment. PhD thesis, University of
Fig. 12. Comparison of VS profile at Site 10 determined from this study to VS Missouri-Columbia, 2008.
profile determined from reflection measurements performed within 800 m of Site [17] Stokoe II KH, Rathje EM, Wilson CR, Rosenblad B. Development of large-scale
10 [33]. Bars on SASW profile indicates range in individual layer velocities to cause mobile shakers and associated instrumentation for in-situ evaluation
10% change in RMS fit to the dispersion curve (dashed lines indicate 420% change of nonlinear characteristics and liquefaction resistance of soils. In: Proceed-
in layer velocity). ings of the 13th World Conference on Earthquake Engineering August
2004.
[18] Rosenblad B, Bertel J. Potential phase unwrapping errors associated with
SASW measurements at soft-over-stiff sites. Geotechnical Testing Journal
reported in these previous studies (  530–587 m/s). The higher 2008;31(5):433–41.
values obtained in this study may be due to the greater depth of [19] Joh S. Advances in interpretation and analysis techniques for spectral-
investigation and the different region of the embayment that was analysis-of surface-waves (SASW) measurements. PhD dissertation, Univer-
sity of Texas at Austin, 1996.
sampled in this study. Other deep VS measurements in the
[20] Roesset JM, Chang D-W, Stokoe KH. Comparison of 2-D and 3-D models for
embayment provide support for the higher Memphis Sand analysis of surface wave tests. In: Proceedings of the 5th international
velocity values obtained in this study. conference on soil dynamics and earthquake engineering. Karlsruhe,
Germany, 1991.
[21] Cramer C, Gomberg J, Schweig E, Waldron B, Tucker K. The Memphis, Shelby
County, Tennessee, seismic hazard maps: US Dept. of the Interior, US
Acknowledgements Geological Survey; 2004. Report no.: 04-1294.
[22] Cramer C, Gomberg J, Schweig ES, Waldron B, Tucker K. First USGS Urban
seismic hazard maps predict the effects of soil. Seismological Research Letters
The authors thank the land owners for allowing access to the 2006;77(1).
field sites, personnel from CERI for assistance in locating [23] Chen K-C, Chiu J-M, Yang Y-T. Qp–Qs relations in the sedimentary basin of
the upper Mississippi embayment using converted phases. Bulletin of the
acceptable sites, and personnel from the NEES site at the Seismological Society of America 1994;84(6):1861–8.
University of Texas at Austin for assistance in the field. This work [24] Menke W. Geophysical data analysis: discrete inverse theory. San Diego:
was supported by the United States Geological Survey (NEHRP Academic Press; 1989.
[25] Tarantola A. Inverse problem theory: methods for data fitting and model
award No. 06-HQGR0050) and the National Science Foundation parameter estimation. New York: Elsevier; 1987.
Network for Earthquake Engineering Simulation program (Grant [26] International Code Council. International Building Code (IBC) 2003.
no. 0530140). [27] Liu H-P, Hu Y, Dorman J, Chang T-S, Chiu J-M. Upper Mississippi embayment
shallow seismic velocities measured in situ. Engineering Geology 1997;
46(3-4):313–30.
[28] Park D, Hashash YMA. Evaluation of seismic site factors in the Mississippi
References embayment. I. Estimation of dynamic properties. Soil Dynamics and Earth-
quake Engineering 2005;25(2):133–44.
[1] Van Arsdale RB, TenBrink RK. Late Cretaceous and Cenozoic geology of the [29] Cramer CH. Quantifying the uncertainty in site amplification modeling and its
New Madrid seismic zone. Bulletin of the Seismological Society of America effects on site-specific seismic-hazard estimation in the upper Mississippi
2000;90(2):345–56. embayment and adjacent areas. Bulletin of the Seismological Society of
[2] Hashash YMA, Park D. Non-linear one-dimensional seismic ground motion America 2006;96(6):2008–20.
propagation in the Mississippi embayment. Engineering Geology [30] Dorman J, Smalley R. Low-frequency seismic surface waves in the
2001;62:185–206. upper Mississippi embayment. Seismological Research Letters 1994;65(2):
[3] Romero S, Rix G. Ground motion amplification of soil in the upper Mississippi 137–148.
embayment: Mid-America Earthquake Center; 2001. Report no.: GIT-CEE/ [31] Williams RA, Wood S, Stephenson WJ, et al. Surface seismic refraction/
GEO-01-1. reflection measurement determinations of potential site resonances and the
ARTICLE IN PRESS
B.L. Rosenblad et al. / Soil Dynamics and Earthquake Engineering 30 (2010) 691–701 701

Areal uniformity of NEHRP site class D in Memphis, Tennessee. Earthquake [33] Street R, Woolery EW. Shear wave velocities of the post-paleozoic sediments
Spectra 2003;19(1):159–89. in the Upper Mississippi Embayment: National Earthquake Hazard Reduction
[32] Street R, Woolery EW, Wang Z, Harris JB. NEHRP soil classifications for Program (NEHRP), Final technical report, Award No. 02HQGR0023, United
estimating site-dependent seismic coefficients in the Upper Mississippi States Geological Survey, 2002.
Embayment. Engineering Geology 2001;62(1-3):123–35.

You might also like