our search engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
Imaging UXO Using Electrical Impedance Tomography

11
JEEG, December 2000, Volume 5, Issue 4, pp. 11-23
Imaging UXO Using Electrical Impedance Tomography
William Daily
1
, Abelardo Ramirez
1
, Robin Newmark
1

and Victor George
2

1
Lawrence Livermore National Laboratory, Livermore, Calif. 94550
2
Roy F. Weston, Vallejo, Calif. 94592
ABSTRACT
This paper reports the results of tests where electrical impedance tomography (EIT) was
evaluated as a tool for detecting and locating buried unexploded ordnance (UXO). The method
relies on the electrolytic polarization induced at the boundary between soil and buried metal. This
induced polarization (IP) produces a measurable phase delay between the electric current imposed
on the subsurface and the resulting voltage distribution. If natural sources of induced polarization
are small compared to those due to buried metal objects, then tomographs of impedance phase
may be used to indicate where metal-soil polarization may be present.
Three controlled tests were performed at a field site containing inert UXO buried in known
locations. These tests produced a phase anomaly of about 20 milliradians that closely matched the
known location of buried UXO objects. A fourth uncontrolled or blind test was performed under a
building without prior knowledge of UXO presence. That test yielded phase anomalies as high as
75 milliradians. Limited excavation was performed at some of these anomalies but only a small
amount (a few tens of grams) of metal was recovered. More extensive excavations are too costly
until the building is razed.
BACKGROUND
Locating UXO is a partially solved problem, meaning that many solutions exist but none work universally.
UXO is typically detected using magnetometers, metal detectors, ground penetrating radar or controlled source
electromagnetic induction. Except for the radar, these techniques provide little information regarding the depth of
burial of potential targets and are sensitive to cultural artifacts such as metal fences, power lines or buildings.
Factors that affect performance of traditional methods include soil moisture content, depth of burial or non-metallic
targets. In some cases, the soil itself generates signals that can confuse the diagnostic method and data
interpretation.
In this work, we evaluate another technology, electrical impedance tomography (EIT, also referred to as
complex resistivity tomography or induced polarization tomography; Oldenberg and Li, 1994; Weller et al., 1996;
Shi et al., 1998; Ramirez et al., 1999), for locating buried UXO. EIT produces a tomograph or image of the
subsurface electrical properties which can provide information regarding the target size, shape and depth. This
information can be useful not only for detecting and locating the target but also for planning and monitoring
remediation. The electrodes required to make EIT measurements can be deployed beneath and around buildings or
other structures, so that surveys can be performed to sense under a building. Most other detection methods are
blinded by interference from the metal in a building. Surveying under buildings is the primary focus of the work
reported herein.
Electrical impedance, long used to probe the subsurface, has two components: magnitude and phase.
Impedance magnitude, or resistivity, defines the distribution of electrical current because flow is concentrated where
resistivity is low and flow is sparse where resistivity is high. Because ordnance composition is typically metallic we
expect buried ordnance to be of low electrical resistivity. Impedance phase, another measure of induced
polarization, describes the behavior of current flow as it depends on frequency (Telford et al., 1976). The metal-
fluid boundary causes a frequency dependant impedance due to the transition from electronic to ionic current flow at
the boundary. Therefore, we expect buried ordnance to produce induced polarization or a phase signal in the
electrical impedance. This polarization arises from the same mechanism responsible for the effect in metallic ore
deposits (Keller and Frischknecht, 1966).
Mare Island, located near Vallejo, California, was a naval base beginning at the Civil War era and its mission
included the manufacture, stockpiling and distribution of naval ordnance (Fig. 1). The Department of the Navy
closed the base in 1996 and plans to return it to civilian control. However, large amounts of buried UXO have been
discovered on site, delaying this transfer. Although some times the material is found scattered, a large accumulation 12


Journal of Environmental and Engineering Geophysics

of UXO is shown in Fig. 2. As a result, the U.S. Navy has supported an active remediation program to render the
site safe for civilian use.
Figure 1. Location of the former Mare Island Naval Shipyard. Numerous areas are currently under
investigation for buried UXO (after SSPORTS, 1997).
The primary method for locating UXO has been magnetic surveys using portable magnetometers (U.S. Army
Report, 1997). The method is fast, economical and easy to use. However, due to interference from ferromagnetic
and ferrimagnetic metal, the magnetometer cannot be used near or under buildings. At Mare Island there are many
buildings which may have been built over buried UXO, and many of these buildings must be removed. Worker
safety could be improved if any UXO could be located and emoved before a building is razed.

13
Daily et al.: Imaging UXO using Electrical Impedance Tomography
Figure 2. Buried mass excavated in dredge ponds at Mare Island. Contents of such masses include welding
rods, silverware, china, tools, and a wide range of UXO.
Electrical Impedance Tomography
Electrical impedance tomography (ETT) is a method that calculates subsurface images of electrical properties
from a large number of impedance measurements. To image under a building, arrays of electrodes are placed
beneath the structure where possible, or around the periphery where that is not possible. A low frequency (perhaps
0.125 to 1.0 Hz) current is driven between two electrodes. As this cur rent flows through the ground under the
building, it establishes voltages at the other electrodes that are measured and recorded. Two other electrodes are
then used to drive current, and voltages are again measured on all other electrodes. This process is repeated until all
linearly independent combinations of current and voltage measurements are made. For 30 electrodes, there are 405
such measurements (n [n 3]/2 where n is the number of electrodes).
The raw data are inverted to produce tomographic images of electrical properties in the ground. For the simpler
case where the impedance is adequately described by the resistance only and the method is called electrical
resistance tomography (ERT) (no phase difference between the current and voltage), the data processing has been de
scribed by Daily and Owen (1991), Oldenburg and Li (1994), Sasaki (1992), and LaBrecque et al. (1996). Early
adaptations of the technique to the field of geophysics were by Pelton et al. (1978), Dines and Lytle (1981), Tripp et
al. (1984), Wexler et al. (1985). Adaptations for medical diagnostics can be found in Isaacson (1986), Barber and
Seager (1987), and Yorkey et al. (1987).
Yang and LaBrecque (1999) describe a three-dimensional inversion algorithm which calculates impedance
magnitude and phase tomographs; this algorithm is used for the work described herein. A two dimensional
algorithm is also used in this work, as described in Ramirez et al. (1999). Other descriptions of EIT algorithms can
be found in Oldenberg and Li (1994); Weller et al. (1996); Yuval and Oldenberg (1997); Shi et al. (1998). Here we
only summarize the general structure of the algorithms used for this work. First, a numerical model of the
subsurface electrical impedance is assumed, for which the voltage field is calculated. These calculated voltages are
compared to those measured; they will be different because the computer model of the subsurface is only an initial
guess. The model is then changed in such a way as to make the voltages calculated for the new model closer to
those measured. The algorithm continues making changes to the numerical model, improving agreement between 14


Journal of Environmental and Engineering Geophysics

calculated and measured voltages. This iterative process is continued until the agreement is within some specified
value that is related to the accuracy of the measured values.
Impeda