| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Hydrology Program, Dep. of Earth & Environmental Science, New Mexico Tech, Socorro, NM 87801
b Dep. of Mathematics, New Mexico Tech, Socorro, NM 87801
* Corresponding author (hendrick{at}nmt.edu)
Received 31 January 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
Abbreviations: GICHD, Geneva International Centre for Humanitarian Demining • GPR, ground penetrating radar
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
No vadose zone contamination is worse than the landmine threat. The United Nations and the U.S. Department of State have declared that landmines are "one of the most widespread, lethal, and long lasting forms of pollution" (Geneva International Centre for Humanitarian Demining [GICHD], 2003). Not only are thousands of persons killed, maimed, and injured each year, but the social, economic, and environmental impacts of landmines are horrendous. In many developing countries the loss of fertile agricultural land and access to water points are among the most serious effects. Rural populations are driven onto increasingly fragile, marginal areas, leading to rapid land degradation (GICHD, 2003) and disturbance of the hydrological cycle.
In practice, current demining techniques involve the use of explosive-sniffing dogs, metal detectors, and mechanical prods. Ground penetrating radar is an alternative technology for landmine detection that has been extensively researched, although it is not yet widely used in practice. Ground penetrating radar has the potential to be much more effective than metal detectors in locating plastic-cased landmines, which have little or no metal content.
Field experiments with GPR have shown that soil conditions can have a large effect on the performance of GPR systems for buried landmine detection. Under some soil conditions the landmine signature is of high quality, while under others no signature can be detected at all. Fritzsche (1995) showed through modeling that GPR signals at 900 MHz would be strongly attenuated in moist soils and especially in clay soils. Trang (1996) found in both simulations and actual experiments with a GPR operating in the 600- to 800-MHz frequency range that it was easier to detect nonmetallic mines when the soil was moist. Johnson and Howard (1999) found that elevated soil moisture actually improves detection by improving the contrast between arid soils and plastic mines at the Energetic Materials Research and Testing Center (New Mexico Tech, Socorro, NM). Scheers et al. (2000) modeled the performance of an ultra wide band GPR operating in the 1- to 5-GHz range for detection of metallic mines and found that the maximum depth at which the mine could be detected decreased as the soil moisture increased. To date, no studies have been conducted that systematically evaluate the effects of soil texture and soil water content on radar signatures from land mines. Therefore, our objectives were (i) to review a suite of models that can be used for the prediction of soil electrical properties and radar responses under a wide range of soil conditions, (ii) to use these models to show the effects that soil texture and water content can have on soil electrical properties, (iii) to conduct field experiments for validation of these models. This study includes plastic and metallic land mines with a wide range of electrical properties. Hence, our results will be applicable not only to landmines but to other buried objects as well.
|
THEORY |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
For this study, we required a model that could predict both the real and imaginary parts of the relative electric permittivity for frequencies near the 900-MHz center frequency of the GPR system that we have used. The available data are the frequency, soil water content, and soil texture. Of the models mentioned above, only the model of Peplinski et al. (1995) satisfies these requirements. This section summarizes the model of Peplinski et al. (1995), which covers the frequency range from 0.3 to 1.3 GHz. This model was based on the Dobson et al. (1985) earlier model for dielectric constants in the 1.4- to 18-GHz range.
In this research we are interested in the attenuation of GPR signals in lossy soils. The static dielectric constant does not adequately represent the frequency-dependent attenuation of GPR signals in these materials. Instead, we will use the complex relative electric permittivity. In the limiting case of a nonlossy soil with no frequency dependence the complex relative electrical permittivity is simply the dielectric constant.
The inputs to the model consist of the volumetric soil water content 
, the frequency f, the fraction of sand particles S, the fraction of clay particles C, the density of the soil particles 
S (a typical value is 2.66 g cm–3), and the dry bulk density of the soil B. An empirically derived formula for effective ionic soil conductivity is the following:
 | [1] |
The sand and clay fractions also enter the model through two empirically derived quantities ß' and ß'', which depend on the soil type but are independent of the frequency and soil water content.
 | [2] |
 | [3] |
Note that in these formulas, S and C are fractions, not percentages. For example, if the clay content is 15%, then C = 0.15.
The real (
'fw) and imaginary (''fw) parts of the complex relative electric permittivity (fw) for the free water are given by a modified Debye model
 | [4] |
 | [5] |
 | [6] |
In these formulas, 0 is the permittivity of free space, w0 is the static dielectric constant of water (80.1 at 20°C), w
is the high-frequency limit of 'fw (4.9 at 20°C), and 
w is the relaxation time of water (9.23 x 10–12 s at 20°C). The dielectric constant of the soil particles (s) is given by the empirical model
 | [7] |
Finally, the real (') and imaginary ('') parts of the complex relative electrical permittivity for the bulk soil are estimated by
 | [8] |
 | [9] |
 | [10] |
The model was fitted to 399 measurements of the real and imaginary parts of the complex relative electrical permittivity of soil samples. The r2 values were 0.985 for the real part (') and 0.940 for the imaginary part ('') (Peplinski et al., 1995).
As GPR signals travel through the soil, the attenuation is controlled by the complex relative electrical permittivity of the soil. The one-way attenuation loss (db) is given by
 | [11] |
is given by
 | [12] |
The models described here have been implemented in a MATLAB package. These MATLAB codes have been made available on the authors' web page at http://www.nmt.edu/~borchers/.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
Sprinkler System
The sprinkler system was built out of PVC tubing (12.7 mm, 1/2 inch) in the shape of a square measuring 3 by 3 m and 1 m tall. Seven Rain Bird (Azusa/Glendora, CA) XS-360TS-1032 spray nozzles were spaced 30 cm apart along the center of the sprinkler system. The system was designed to wet our experimental sites in a uniform manner. For details we refer the reader to Miller (2002).
Antitank Landmines
The simulated landmines used at the Socorro site were completely inert and composed of Dow Corning (Midland, MI) 3110 RTV Silicon Rubber. They are designed to simulate the NR26, an antitank landmine, which is a nonmetallic landmine and has dimensions of 30.0 cm in diameter and 11.5 cm in height. TNO Physics and Electronics Laboratory in the Netherlands manufactured these simulated landmines.
Real antitank landmines were used at the Yuma Proving Ground site. These landmines have been defused for safety, but still contain their explosive charges.
Ground Penetrating Radar System
The measurements described in this study were performed with a Sensors & Software (Mississauga, ON, Canada) Pulse EKKO 1000 GPR system. The system was operated with 900-MHz antennas. This puts our measurements well within the 0.3- to 1.3-GHz band considered in the model of Peplinski et al. (1995). To ensure consistent antenna location, a wooden frame was used to position the antennas. This frame ensured that the horizontal positioning of the antennas was consistent for each experiment. The frame positioned the antennas approximately 4 cm above the ground.
In practice, a commercial GPR of this particular type would not be used for landmine detection. Most of the systems that have been field tested make use of radar in a "look ahead" configuration. However, we were limited in this research by the capabilities of the available GPR system.
Signal Processing Techniques
Seismic Unix was used for all the post data collection image processing. Seismic Unix is a signal processing software package developed by the Center for Wave Phenomena at the Colorado School of Mines (Stockwell and Cohen, 2001). A zero-phase, sine-squared tapered filter was applied to each image in the form of a highpass filter. Significant ground bounce appeared in the upper part of each GPR profile. This was caused by the signal bouncing off the ground and ringing between the transmitting and receiving antennas. To delete this noise from the images the amplitude for this portion of the traces was set to zero.
Field Sites
Three field sites were chosen based on their soil texture: sand, silt, and clay. The sand and silt sites were located in the Sevilleta National Wildlife Refuge 20 km north of Socorro, NM, and the clay site was located in the Bosque Del Apache National Wildlife Refuge 40 km south of Socorro, NM. The sand soil had a composition of 95% sand, 2% silt, and 3% clay, and a dry bulk density of 1.60 g cm–3. The silt soil had a composition of 2% sand, 66% silt, and 32% clay, and a dry bulk density of 1.30 g cm–3. The clay site was located on the floodplain of the Rio Grande River in the Bosque Del Apache Refuge. This soil had a textural composition of 1% sand, 27% silt, and 72% clay, and a dry bulk density of 1.54 g cm–3. Using the U.S. Department of Agriculture classification scheme (Klute, 1986), the sand site soil was classified as a sand, the silt site soil was classified as a silty clay loam, and the clay site soil as a clay.
The U.S. Army's Yuma Proving Ground is located near the Arizona–California border, adjacent to the Colorado River, approximately 24 miles north of the city of Yuma, AZ. The Countermine Testing and Training Range is located in the Kofa region of the area. In the Countermine Testing and Training Range there are two types of landmine lanes: the Handheld Detector Mine Lanes and the Vehicle-Mounted Detector Mine Lanes. Both of these lanes have a mixture of nonmetallic and metallic, foreign and domestic defused antitank landmines. The Yuma Handheld Detection Mine Lane has a soil composition of 80% sand, 14% silt, and 6% clay and is classified by the USDA classification scheme as loamy sand. The Vehicle-Mounted Detector Mine Lane has a soil composition of 57% sand, 28% silt, and 15% clay and is classified as a sandy loam.
The following describes the general procedures used for burying the landmines at the sites in the Socorro, NM area. First, a 3- by 3-m plot was cleared of any grass, shrubs, or other obstacles. Then the soil surface was leveled so that the surface was flat without any sloping edges. Inside this area an antitank landmine was buried 11 cm deep. Then a second antitank landmine was buried approximately 1.5 m away from the first landmine also 11 cm deep. This second landmine had TDR probes buried above and below to measure the soil moisture around the landmine. The TDR probes were buried at 3, 8, 23, and 28 cm below the ground surface. In this study we assumed that the water content distribution measured around the mine instrumented with the TDR probes is equal to that around the mine without TDR probes. At the latter mine we measured radar responses without interference from the TDR probes.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
Dielectric Constant vs. Soil Water Content Predictions
Figure 1
shows how the complex relative electrical permittivity changes with soil water content for the three soils from the Socorro, NM area. In this figure the predicted real part (solid lines) increases as the soil water content is raised, where the imaginary part (dotted lines) remains almost constant for the entire range of soil water contents. In this section, all model predictions are given for a frequency of 900 MHz, since that is the operating frequency of the GPR used in this study. The water contents used in the model are those measured in the field before and after water application with the sprinkler system (Miller, 2002).
|
If a nonporous plastic landmine is buried in a sand, silt, or clay soil, then as the soil water content increases, the bulk dielectric constant of the soil also increases, while the dielectric constant of the landmine remains the same (about 3). This elevation in the dielectric constant of the bulk soil will lead to a larger reflection coefficient (approaching unity), which in theory will lead to an improved image of the landmine. If the bulk dielectric constant and soil water content are the only factors examined, one may come to the erroneous conclusion that in all soils landmine detection will improve with increasing soil water content because the dielectric constant contrast increases with elevated soil water contents. In the next two sections we explain the roles frequency and attenuation can have on the complex relative electrical permittivity.
Dielectric Constant vs. Frequency Predictions
The complex relative dielectric permittivity of a soil also changes as a function of the frequency of the radar waves. Figure 2 shows how the complex relative electrical permittivity varies with frequency for the same Socorro soils at dry and wet soil water conditions.
|
For the Sevilleta silt soil, the imaginary part of the complex relative electrical permittivity decreases significantly for the 0.3- to 1.3-GHz range for both the dry and wet soil water conditions.
For the Bosque clay soil at the low frequency range (0.3–1.3 GHz), the imaginary part or loss term is extremely significant when the soil is wet, changing by 7 at this range. When the soil is dry, the imaginary part of the dielectric constant decreases but is not as significant (changing by 1 at this range) as when the soil is wet. High clay content plays a significant role in elevating the imaginary part, as seen in the Bosque clay and Sevilleta silt soils.
Attenuation and Radar Response
From Eq. [11] and [12] it is clear that radar wave attenuation should increase as the frequency of the radar increases and as the ratio of the imaginary part of the dielectric constant to real part increases. This ratio of imaginary to real part will generally increase as the soil water content is increased at a given frequency. Figure 3
shows the predictions for the attenuation at a range of soil water contents for the three Socorro soils at 900 MHz. From this figure, it is obvious that as the clay content of the soil increases, so does the amount of attenuation at a given soil water content. The Sevilleta sand soil (solid line) will attenuate about 20 db m–1, where the Sevilleta silt soil (dashed line) will attenuate about 50 db m–1, and the Bosque clay soil (dotted line) will attenuate about 65 db m–1 at 40% soil water content. Figure 4 shows how changes in frequency relate to radar wave attenuation. As frequency increases, the attenuation of the GPR signal in sand, silt, and clay soils increases rapidly.
|
|
|
In interpreting these graphs, it is also important to note that the vertical axis is travel time (in ns), not depth. Conversion to depth requires knowledge of the locations of the GPR antennas and the velocity of the GPR signal in the soil. This can be difficult because neither factor is known precisely. In practice, we were not able to precisely position the antennas, so travel times and depths are not strictly comparable between our experiments. For further discussion of these issues, see Miller (2002). We can generally expect that the velocity of the GPR signal in the soil will decrease as water content and the real part of the dielectric constant increase. Since the velocity of the GPR signal decreases under these circumstances, the travel time typically increases. This pattern is seen in most of the following results.
Socorro, NM Test Sites—Simulated Nonmetallic Landmines
Figure 6
shows a series of profiles of buried landmines in the three Socorro soils. The first image is a profile of the simulated landmine when the volumetric soil water content above the landmine is at 7%. The hyperbolic feature seen between the 20th and the 40th trace indicates the buried simulated landmine. The profile to the right of this is the same site after the volumetric soil water content above the landmine was raised to 29%. The landmine is again indicated by the hyperbolic feature and seen directly under the 20th trace mark. These profiles clearly demonstrate that raising the volumetric soil water content of dry sandy soils can enhance the ability of the GPR to image landmines, which is in agreement with what our model predicts.
|
Applying water to dry clay soils, however, does not enhance detection. The third row in Fig. 6 shows two GPR wiggle trace plots from the Bosque Del Apache clay soil site. The first is an image taken during dry field conditions, with 5% volumetric soil water content above the landmine. The landmine is detectable under the dry clay soil conditions shown in this figure. The hyperbolic feature directly below the 35th trace mark on the horizontal scale indicates the location of the landmine in this image. The second profile shows an image of the same Bosque clay soil after infiltrating a total of 2700 L of water, raising the volumetric soil water content to 42% around the landmine. After application of large amounts of water, the landmine is clearly invisible to GPR. This is expected due to the extremely large attenuation in the wet clay, as our model suggests.
Yuma, AZ Test Sites—Nonmetallic Landmines
In this section the results from the Yuma Proving Ground landmine test lanes are presented. The four profiles seen in Fig. 7
show GPR images of buried nonmetallic antitank landmines from the Handheld test range under both dry and wet soil conditions. The first profile in Fig. 7 is a wiggle trace plot of a VS–1.6 antitank landmine buried 7.62 cm deep in dry loamy sand soil. The VS–1.6 is a low metal antitank landmine and contains a high explosive main charge with a surrogate RTV-3110 silicon rubber booster. The detonator shaft is the only metallic component of the landmine. In this profile a small reflection from the top of the landmine can be seen at the 34th trace. The contrast in the dielectric constant between the landmine and the surrounding soil is not large enough to produce a significant reflection, so detection is difficult. The second profile is an image of the same landmine after the soil water content was raised to 26%. A stronger reflection is produced from the landmine and detection is enhanced through watering the soil.
|
Yuma, AZ Test Sites—Metallic Landmines
Figure 8
presents GPR images of two metallic antitank landmines. The first image in the upper row is of a TM62M metallic landmine buried in loamy sand soil. This figure shows a very strong reflection from the landmine when the soil is dry, since metallic landmines should produce perfect reflection because their reflection coefficients are equal to unity. The second profile is a radar image of the same landmine after the soil water content was raised to 26% above the landmine. The landmine in this figure would most likely not be detectable with GPR at greater soil water contents. Metallic landmines have dielectric constants that are very large, approaching infinity, so the contrast between these types of landmines and the soil is also very large, which should always produce significant reflections. However, as seen in this example, applying water in certain situations does not enhance detection, rather it produces the opposite effect.
|
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
In sand and silt soils, the Peplinski model predicts that at 900 MHz the real part of the complex relative electrical permittivity will increase rapidly as the soil water content is increased. Since the dielectric constant of the mine remains constant, the contrast between the mine and the soil will also increase rapidly. This suggests that for these soils at elevated soil water conditions, there will be enough dielectric contrast to detect landmines. In addition, the total attenuation for these types of soils is relatively low, so it will not hinder detection of landmines.
In clay soils at 900 MHz, the real part of the complex relative electrical permittivity increases rapidly as the soil water content is increased from dry soil to wet soil. However, the total attenuation in clay soils is very large, approaching 65 db m–1 at 40% soil water content. This suggests that landmine detection will not improve at elevated soil water conditions in this type of soil due to the strong attenuation.
Ground penetrating radar profiles of buried simulated nonmetallic antitank landmines in sand and silt soils at 900 MHz become clearer as the soil water content is increased from dry to wet. Ground penetrating radar profiles of buried simulated nonmetallic antitank landmines in clay soils at 900 MHz do not become clearer as the soil water content is increased from dry to wet. Ground penetrating radar profiles of buried metallic antitank landmines in sand and silt soils at 900 MHz do not become clearer as the soil water content is increased from dry to wet.
This study of the physics of landmine–radar–soil systems demonstrates both the great potential and the pitfalls of landmine sensors based on GPR. Radar works well with nonmetallic mines in wet sand and silt soils and in dry clay soils whereas metallic mines are best detected in dry soils. Unfortunately, soil texture (e.g., Hendrickx et al., 1986; Wierenga et al., 1987) and water content (e.g., Hendrickx et al., 1990, 1993; Jaramillo et al., 2000; Yao and Hendrickx, 2001) can change in relatively short distances. Soil water content distributions around landmines exhibit a large temporal variability (Das et al., 2001; Rhebergen et al., 2002). Hendrickx et al. (2001) and Lensen et al. (2001) demonstrated how the spatial variability of electrical soil properties is caused by changes in soil texture and water content. These factors should be considered before and during deployment of a GPR system for landmine detection.
Hendrickx et al. (2003) discussed how worldwide soil databases can be used to predict soil electrical properties, yet there is no simple prescription for when GPR will be effective. Rather, the model discussed in this paper should be used to determine the likely dielectric contrast between mine and soil and the likely attenuation of the GPR signal. In conditions where the dielectric contrast is strong and attenuation is mild, GPR is likely to work well. In other cases, it might be possible to alter the soil water content so as to improve the dielectric contrast while keeping the attenuation at an acceptable level.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONTHEORYMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARYREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Vereecken, S. Hubbard, A. Binley, and T. Ferre Hydrogeophysics: An Introduction from the Guest Editors Vadose Zone J., November 1, 2004; 3(4): 1060 - 1062. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
