Post-processing and display of data
of the targets known to be buried at JPG and because it
predicts the correct phase polarity. For a metal target,
We first band-pass-filtered (very wide settings, e.g.,
50600 MHz for the Model 5103 antenna) the recorded
higher than εs* and produces a wavelet with a phase
data to alleviate high-frequency electronic noise and
structure opposite to that produced when εt* is lower
low-frequency, above-surface clutter. We normalized
than εs*.
the number of data traces between event markers over
the targets we emplaced to compensate for changes in
It is unlikely that any geologic or organic inhomo-
geneity in the JPG soil we profiled had a higher εt*
dragging speed. We did not normalize the longer pro-
files with the 100-ft (30-m) marker spacing because
than that of the soil itself. Consistent horizons are vir-
vehicle speed varied between any two markers.
tually absent in our data, which means that electrically
We used both linear and nonlinear gray-scale for-
important changes, such as in moisture content, were
gradational. In addition, ε′of limestone is generally
mats to indicate signal strength, and used an amplitude
format to display the profiles for the targets we buried.
between 8 and 10 (Parkhomenko 1967), which is near
In this format, positive phase is indicated by lighter
that of the soil and precluded strong bedrock reflections.
tones and negative phase by darker tones. We used an
intensity format to display the profiles of the perma-
RESULTS AND DISCUSSION
nent targets at JPG. In this format, which is insensitive
to phase, strength is indicated by the intensity of darker
Control studies
tones.
Our objectives for the control studies were to obtain
Profile interpretation
profile responses and scattered waveforms for buried
The main objectives of the profile analysis are to
metal reflectors, as well as soil moisture and conduc-
determine if ordnance targets had been detected and
tivity profiles. We conducted these studies either out-
the range of ε′values for the site soil. The permittivity
side or along the perimeter of the 40-acre site (Fig. 3).
analysis used the diffractions caused by radar scatter-
We buried two 9-inch- (23-cm-) diameter metal disks
ing from targets. In this method we matched the hyper-
at depths of 11 (28 cm) and 23 (58 cm) inches. The
bolic shape of the diffractions with theoretical hyper-
removed soil was highly compact and did not appear to
bolas for a given value of ε′ (Jezek et al. 1979, Clarke
have excess moisture. Therefore, we think that no sig-
and Bentley 1994, Arcone et al. 1998). The main dis-
nificant soil drying took place between removal and
reburial. The 300- and 600-MHz diffractions from the
advantages of this approach are 1) the hyperbolas can
deeper target (Fig. 4) (the response to the more shallow
actually be responses to linear soil inclusions, in which
target is not sufficiently separated from the direct cou-
case the hyperbolas are distorted reflections that result
pling between antennas to facilitate analysis) best fit
when the transect obliquely intersects the inclusion
theoretical diffraction hyperbolas for ε′ = 9.3 and 8.6 at
direction (Jezek et al. 1979) and are thus artificially
300 and 600 MHz, respectively. The values of ε′, which
wide; and 2) an erratic towing speed, which would dis-
we computed from the wavelet round-trip travel time
tort the hyperbolic image. Item 1 was not considered
when the antennas were over the center of the targets,
important because of the depositional process of the
are 9.5 and 8.7, respectively. In accordance with the
soil (glacial drift and loess) and because of probable
measurements, dielectric dispersion theory (eq. 4, and
historical tilling. Item 2 is a concern and for this reason
discussed below) predicts that the 600-MHz value
a statistical study is presented.
should be slightly less than the 300-MHz value.
Target detection depends on the presence of either
The accompanying traces in Figure 4, whose posi-
or both diffractions and reflections and also on their
tions within the profiles are indicated by arrows, show
phase polarity. Both the strength and phase polarity of
the forms of the scattered wavelets within the diffrac-
a reflected or diffracted event depend on the reflectivity
tions. The wavelets have a negativepositivenegative
of a target, which is determined by its Fresnel reflec-
sequence to the phase polarity of the dominant half-
tion coefficient, R, such that
cycles. This sequence is typical for the relative polarity
R = (εs*1/2 εt*1/2)/( εs*1/2 + εt*1/2)
(6)
wiring of these GSSI antennas and is characteristic of
targets whose wave impedance (eq. 6) is higher than
where εt* is the complex permittivity for the target
that of the surrounding media. Targets characterized by
an ε′ value less than that of the soil matrix would pro-
medium (Wait 1970). Although this formula applies to
plane wave incidence upon large flat reflectors, we
duce a similar wavelet but with opposite phase polarity
invoke its use because of the small in-situ wavelengths
of the individual half-cycles. The local frequency is
(30 cm at 300 MHz) relative to the larger sizes of some
indicated for the wavelets.
5
Previous Page
