APPENDIX A: GROUND-PENETRATING RADAR OPERATION AND THEORY
records between distance markers were equal-
General operation
A Geophysical Survey Systems, Inc., SIR Model
ized (i.e., "justified") to compensate for changes
4800 radar control unit, assorted antennas, cables,
in towing speed. Later processing reduced back-
a GSSI DT6000 digital tape recorder, and a power
ground ringing (e.g., from surface meltwater), al-
supply were used in this study. The control unit
leviated overall gain variations resulting from
keys the transmitter on and off at 50 kHz (syn-
inconsistent antennaground coupling, and en-
chronized with the receiver), sets the scan rate
hanced vertical resolution using deconvolution.
(rate at which echo scans are compiled, generally
Data were displayed as echo-time-of-return vs.
25.6 scans/s), scan time range and the time range
profile distance using either wiggle trace or line
gain to be applied to the scans. The transmit an-
intensity format. Individual scans are easily re-
tenna radiates a broadband pulse that lasts from
trieved with the software.
only a few to tens of nanoseconds (ns). A sepa-
Data interpretation is generally based on the
rate, identical receive antenna is employed be-
simple echo delay formula
cause echoes can return from near-surface targets
before the transmit antenna has stopped radiat-
d = ct/2n
(A1)
ing. The received signals are converted by sam-
where d is the depth of a reflector in centimeters, t
ing, amplification and recording.
is the echo time delay in nanoseconds and c is the
speed of electromagnetic waves in a vacuum (30
cm/ns). The quantity n is often replaced by ε1/2,
Field procedures
where ε is the dielectric constant (approximately
Transects were established along trails in the
Canol Road area and profiled by dragging the
5.36 for frozen alluvium). The factor of 2 in eq
antennas along the ground, about 67 m behind a
A1 accounts for the round-trip propagation path
tracked vehicle. All surveys were done with the
of the pulse and applies only to reflections from
antennas polarized perpendicular to the transect
horizontally flat interfaces extending several in-
direction so that the more intense, but unortho-
situ wavelengths, or to scattering from point
dox, H-plane pattern was in the line of the transect.
sources. Equation A1 can be applied to several
Event markers were placed artificially on the
layers successively if n is known for each layer
records at measured intervals, generally every 10
and the delay times to each layer interface are
m. Scan durations were set to encompass the fea-
easily picked off the record. Measurements made
tures of interest, which always occurred within
during the summer, when conditions had re-
the depth of the noise floor visible on the oscillo-
mained dry for over 6 weeks, reveal that a round-
trip propagation time of 100 ns corresponds to a
scope incorporated in the control unit. Generally,
depth of about 6.5 m in permafrost, and 24 ns
we used 500 to 1000 ns for profiles with both the
corresponds to 1 m of surface thaw.
50- and 100-MHz antennas as the deeper sections
Proper translation of echo time of return into
of water beneath permafrost usually occurred be-
depth requires a complicated process known as
tween 400 and 600 ns into the records. Scan dura-
migration. Successful migration requires detailed
tions up to 2000 ns for the 50-MHz antennas were
knowledge of the subsurface velocity structure.
used in areas with much deeper (to 60 m) perma-
Our knowledge of permafrost and thaw velocity
frost.
comes from correlations between GPR and bore-
hole data, along with previous studies of refrac-
Data recording
tion in the Canol area. Migration is not necessary
and processing
when subsurface interfaces are fairly horizontal,
All data were recorded at a scan rate of 25.6/s,
as they are in this study. In viewing the GPR
and a scan sampling density of 512 or 1024 eight-
profiles it must be remembered that there is a
bit words. The slow dragging speed resulted in a
very large compression in the horizontal direc-
recorded scan about every 4 cm. All data were
tion.
stacked five-fold, which marginally affected de-
tail as the stacked horizontal data density was
about one scan per 20 cm. Data recorded when
Antennas and radiation
All antennas used are resistively loaded di-
antenna movement was stalled were deleted from
poles that radiate a pulse containing about 2 to
the total record. After editing, the lengths of the
23
Previous Page
