of 1.013 with a variance of 0.05711. In Figure 5d the
3. FIELD DATA ACQUISITION AND
phase estimate population yields a mean of 39.75
EXPERIMENTAL CONDITIONS
with a variance of 3.2492. The critical point to note
To test the efficacy of beamforming, experimen-
is the decrease in the variance for the case of the
tal measurements of various sources were ob-
overlapped six block estimates as compared with
tained. The environmental characteristics of the
the variance calculated for the single block esti-
test site and the methods used to make these
mates. Thus, it is statistically advantageous to use
recordings are discussed in this section.
OBAFFT to increase the stability of and decrease
the bias in spectral estimation.
3.1. Geological setting
The field tests were performed in February 1988
2.4. Verification of signal-processing algorithm
at Range 37 at the Michigan National Guard's
The Fortran code implementing the beamform-
Camp Grayling (Fig. 6). The soil in the Grayling
ing procedures given above is discussed in detail
locale consists of glacial outwash sand and gravel
by Moran (1991). The algorithm was extensively
(Farrland and Bell 1982). Logs from four wells in
tested during development and continues to be
the vicinity show the thickness of these deposits at
refined. Two tests were used to validate the proce-
116, 137, 175, and 251 m, with the bedrock below
dure. The first test was numerical, using a wide
being Mississippian in age (Lilienthal 1978). A
variety of synthetic signals under various signal-
local driller estimated the sand thickness in the
to-noise ratios (SNRs) and with a variety of fre-
area to be around 180 m based on his experience.
quency bands and array geometries. The second
A borehole 30 m deep was drilled at the site; the
and most critical test came from results obtained
soils logged during the drilling are given in Table
from field data collected under defined experi-
1. The characteristics of eight soil samples that
mental conditions. These field data tests are based
were collected off the drill auger are listed in Table
on the impulsive sources discussed in section 5.2.
2. These observations show that the site was un-
derlain by medium sand, with some thin layers of
fine gravel and clay at depths around 4 and 24 m.
The water table was 5.8 m below the surface.
3.2. Shallow seismic velocity structure
Standard compressional (P) and shear (S) wave
refraction methods were applied at the site to
N
determine the seismic velocity structure. A linear
SMet
array of 24 geophones spaced 1 m apart was ar-
tation
Vehicle
A
ranged perpendicular to the test track. Hammer
rray
blows were then recorded at intervals off each end
ReAraction
f
rray
of the array. The resulting distance vs. travel time
Recording
T
curves were analyzed using the intercept time
k
rac
railer
T
method to determine the seismic velocity as a
st
Te
le
hic
function of depth. For the P waves, vertical sledge-
Ve
hammer blows on a metal plate at the ground
surface were used as the source, and Mark Prod-
ucts Model L15-B vertical-component geophones
with resonant frequencies of 4.5 Hz were used to
record the resulting ground motion. For the S
waves, a 0.2 0.2 m wooden plank was aligned
perpendicular to the array and clamped to the
ground by the front wheels of a pickup truck (Fig.
7). Horizontal sledgehammer blows on each end
of the plank, which excited SH (shear horizontal)
Figure 6. Location map for the seismic experiments,
waves of opposite polarity, were then recorded at
showing vehicle test track, location of contractor's sen-
each source location using horizontal-component
sors, and location of CRREL seismic arrays. Also shown
geophones.
is the location of the snow pits, met station, frost tubes,
One difficulty encountered was caused by the
and borehole.
9
Previous Page
