Ž .
Journal of Applied Geophysics 43 2000 175–187 www.elsevier.nlrlocaterjappgeo
Maximizing the information return from ground penetrating radar
Gary R. Olhoeft
Department of Geophysics, Colorado School of Mines, Golden, CO 80401-1887 USA Received 2 February 1999; received in revised form 23 April 1999; accepted 24 May 1999
Abstract
Ground penetrating radar data is not always easy to acquire, and sometimes the acquisition may be constrained by equipment availability, weather, legal or logistical constraints, safety or access considerations. Examples of these include
archaeological or geotechnical sites about to be excavated, contaminated lands undergoing remediation, hazardous areas such as unexploded ordnance lands or active volcanoes, and difficult to visit locations such as Antarctica or the surface of Mars.
These situations may result in only one chance at acquiring data. Thus, the data need to be acquired, processed and modeled with the aim of maximizing the information return for the time, cost and hazard risked. This process begins by properly
setting up the survey with the expectation of the site conditions but allowing for flexibility and serendipity in the unknown. Not only are radar data acquired, but also calibration, orientation, location and other required parameters describing the
equipment and survey are recorded. All of these parameters are used in the processing and modeling of the data. The final results will be not just a radar image as a pseudo-cross-section, but a corrected geometric cross-section, interpreted electrical
and magnetic properties of the ground, location, orientation, size and shape of subsurface objects, and composition of the ground and objects as inferred density, porosity, fluid saturation, and other relevant material occurrence properties. q 2000
Elsevier Science B.V. All rights reserved.
Keywords: Acquisition; Processing; Modeling; Interpretation; Display; Utility detection
1. Introduction
For most of the history of ground penetrating radar, the instruments have been used to acquire
data that have been presented as distorted im- ages or pseudo-cross-sections of the subsurface
Ž .
Morey, 1974; Olhoeft, 1988 . Such images of- ten solve problems such as the horizontal loca-
tion of some change or thing buried, without any further necessity of processing. However,
many problems pose questions requiring an- swers with more detailed information, such as
Fax: q1-303-273-3478; e-mail: golhoeftmines.edu
what is the depth to a buried utility pipe and its size, orientation, and composition? What are the
depth, size, shape and orientation of the unex- ploded ordnance? What is the density of com-
paction of soil in the bridge approach? How is the fluid saturation of contaminant perched on
this clay layer changing with time as the site is remediated? These questions require quantita-
tive answers only obtained by properly acquir- ing the radar data, processing and modeling the
data, and interpreting the results, including a display in terms the person dealing with the
problem can understand. The physical processes and the equations to describe them have been
0926-9851r00r - see front matter q 2000 Elsevier Science B.V. All rights reserved. Ž
. PII: S 0 9 2 6 - 9 8 5 1 9 9 0 0 0 5 7 - 9
known for a long time. Only in the past decade have they been applied to answer some of the
questions above, but they are still not routinely applied.
2. Locating a buried pipe
Fig. 1 illustrates a portion of a radar pseudo- section acquired with a GSSI SIR-10A q radar
Ž
1
. system using a 900-MHz in air
center fre- quency bistatic bow-tie antenna over a 90-cm
Ž diameter steel pipe buried 37 cm deep surface
. Ž
to top of pipe near Yuma, AZ Olhoeft et al., .
1994 . The radar reflection from the pipe is the result of electromagnetic wave propagation de-
scribed by the radar equation and geometry through, among other things, the Fresnel reflec-
tion coefficient in amplitude, Snell’s law in angle, and the Stokes scattering matrix in polar-
Ž ization
Balanis, 1989; Powers, 1995; Smith, .
1997 . The location and orientation of the pipe were known, so the electric fields of the anten-
nas were set horizontal, parallel to each other, parallel to the long axis of the pipe, and to
traverse perpendicular across the pipe, maximiz- ing the coupling with respect to polarization.
Ž Across the top of the image are marks small
. white vertical bars used to locate the antenna
and later correct for variations of towing speed. Each mark indicates passage of the center of the
antenna past a flag in a series spaced 1 m apart along the antenna traverse path. The vertical
black line running through the middle of the image is the location of the single scan wiggle
trace plotted to the right. The two way travel time vertical scale is also shown as an equiva-
lent depth assuming a relative dielectric permit- tivity of four. The horizontal line across the top
of the image currently displays nothing and will
1
Most GPR antennas are ground-loaded, lowering their center frequency when in contact with the ground by an
amount determined by the antenna design and the electro- magnetic properties of the ground.
Fig. 1. Raw radar data acquired with a GSSI SIR-10Aq Ž
. using a bistatic 900 MHz in air center frequency antenna
towed across a 90-cm diameter pipe buried 37 cm deep.
be explained in later figures when it shows features in the data.
The data in the radar image exhibit several problems. The image is horizontally distorted
Ž by uneven towing speed uneven spacing of the
. marks across the top and vertically by an un-
known velocity of wave propagation. There is horizontal banding running across the image
from less than optimal coupling of the antenna to the ground and unwanted oscillatory ringing
of the antenna. There are five vertical lines coming up from the bottom of the image caused
by radio frequency interference from nearby
Ž portable radios or cell phones see Appendix
. A . Nonetheless, about a third of the way down
from the top, a reflection caused by a layer in the geology may be seen to run horizontally
across the image, broken in the middle by the trench created to bury the pipe, and exhibiting
the characteristic ‘‘hyperbola’’ scattering shape caused by the metal pipe. If the problem were
utility location, then the problem is solved at this point by noting the presence of a metallic
reflector at the horizontal position of the top of the hyperbola. By rotating the antenna electric
field orientation while centered above the pipe location indicated by the hyperbola, the az-
imuthal strike orientation of the pipe may also be quickly determined from the change in polar-
ization response. By measuring the change in polarization with different antenna geometries
Žoften called HH, VV and HV polarizations;
. van Zyl and Ulaby, 1990 and solving for the
Stokes–Mueller polarization matrices, both the strike and dip of the pipe may be determined.
For the remainder of this example, polarization will be ignored as the data acquisition was setup
to be maximally coupled between the antenna electric field and the long axis of the pipe.
3. Estimating the pipe depth and size