Version Date:  21 September 1999

Landmine Detection:  The Problem and the Challenge
For a shorter version of this article see: Applied Radiation and Isotopes, Volume 53, pp. 557-563, 2000

Esam M.A. Hussein (Hussein@unb.ca)
and
Edward J. Waller (Waller@unb.ca)

Laboratory for Threat Material Detection
http://www.unb.ca/ME/LTMD/
Department of Mechanical Engineering
University of New Brunswick
P.O. Box 4400, Fredericton, NB, Canada, E3B 5A3

In December 1997, 123 countries signed the "Convention on the Prohibition of the Use, Stockpiling, Production and Transfer of Anti-Personnel Mines and on Their Destruction" in Ottawa, Canada. The convention requested, among other things, that "Each State Party in a position to do so shall provide assistance for mine clearance and related activities". In recognition of the inabilities of some countries to do so, the Convention also stated that "States Parties may request the United Nations, regional organizations, other States Parties or other competent intergovernmental or non-governmental fora to assist its authorities in the elaboration of a national demining program".
     Naturally, the first step in the demining process is to locate the mines. Maps, depicting the location and layout of minefields, are rarely available. Therefore, detection by an opposing force, a third party contingent, or even the force responsible for the laying of the minefield, becomes essential. The detection process is generally conducted in two stages: first on a global scale to locate a potential minefield and target anomalies in the ground, and second to locally determine the character of a particular anomaly. Both global searching and local identification present challenges to detection technologies. A global search must be rapid and accurate for targeting anomalies in the ground, whereas local search must be accurate for detecting landmines or, more precisely, explosives, in the ground.
     The purpose of this paper is to present the challenges, caused by the variety of mines, detection requirements and terrain, facing local area landmine detection technologies. The paper also assesses the suitability of different methods for mine detection and the role radiation and radioisotopes can play in helping humanity get rid of this scourge. First, however, a brief description of the significance and extent of the threat presented by landmines is given.

Landmines, since the First World War, have proved to be an effective military weapon. They deny an enemy access to areas, and can deflect, delay and/or destroy enemy forces. They can also serve as a nuisance to disrupt lines of supply and demoralize opposing forces. Antitank (AT) mines disrupt vehicular traffic, while antipersonnel (AP) mines protect antitank mines, defend areas and deny access to bridges and other assets. Under some circumstances, landmines are used to control military and civilian movement across borders. In guerrilla and civil wars, landmines are used to disrupt commerce, instill fear among non-combatants, as a psychological weapon that undermines confidence in the local government, and as booby traps. Although landmines are seen as an effective and inexpensive weapon, they represent a threat to public safety, i.e. to innocent bystanders and civilians. They undermine peace and stability and leave behind maimed individuals who require continuing health care and may cease to be fully productive members of society.
     It is estimated that there are from 50 to 70 million uncleared mines within at least 70 countries, with a clearance rate that lags the emplacement rate. About 26,000 people are killed or maimed every year by landmines. For example, in Angola one of every 334 individuals is a landmine amputee, and Cambodia has greater than 25,000 amputees due to mine blasts. The lives of over 22 million people are impeded from return to normalcy by landmines. The US Department of State (1998) provides further information on the global landmine crisis.

Antitank and antipersonnel mines come in all shapes and sizes, and can be encased in metal, plastic, wood or nothing at all. Their fusing mechanism varies from simple pressure triggers to trip wires, tilt rods, acoustic and seismic fuses, or even light- or magnetic-influenced fuses. They can be embedded in a field cluttered with various materials and objects, buried underground at various depths, scattered on the surface, planted within buildings, or covered by plant overgrowth.
     Antitank or Antivehicular (AT) land mines often have the shape of truncated cylinders or squares with round corners, with a largest dimension from 150 to 300 mm, and a thickness of 50 to 90 mm. The explosive material is typically TNT, Comp B, or RDX. AT's are buried at various depths from flush to the surface to greater than 150 mm (MineFacts, 1995). AT mines are usually associated with warfare, and confined to battle fields, which can be corridored thus minimizing the risk to the general public.
     Anti-personnel (AP) land mines and booby traps are typically shaped in the form of a disk or a cylinder, with diameters from 20 to 125 mm, length from 50 to 100 mm, and can weigh as little as 30 g. TNT, Tetryl, and Comp B are the common types of explosives used. AP's are usually shallow buried, at a range from flush to the surface to a maximum depth of about 50 mm; as they do reduced damage if buried any deeper (MineFacts, 1995). AP mines pose a greater danger to civilians, as they are often used in civil and guerrilla wars and placed in areas accessible to the public.

The requirements of civilian demining (mine clearance) are quite different from those of military demining; and this affects the detection problem. During a military countermine operation, the objective is to breach a minefield as fast as possible, often using brute force. The demining operation then involves identifying a minefield and breaching it using explosive charges, flails, rollers, plows or rakes. The objective is clearing a path for crossing, typically one vehicle wide and as long as required, with limited casualties. Plows and rakes leave berms containing unexploded mines that are often ignored and left as a problem for others. In battles, a minefield can also be marked and avoided altogether by going around it. Simply put, in military demining one is not concerned with problems that can be addressed at a later time.
     Civilian demining, on the other hand, is more difficult and dangerous than military demining, as it requires complete removal of all mines. Normal traffic and use of land must be re-established and therefore absolutely no explosive material can be left un-removed. This is also necessary to restore public confidence, since even rumours can lend an entire field or road useless.
     Post-cleaning of cleared roads or fields, also called proofing, is essential to rebuild the public's confidence. This process involves the detection and removal of any mines that may remain undetected. Mines left undetected in the first phase of demining are likely those that were more deeply buried and can cause future problems if left undetected. These aspects make the detection process even more challenging, and may require special proofing detection technology.

A landmine detection system should be able to detect mines regardless of the type of explosives used, since mines are made of a variety of explosive materials. Mines come in a variety of shapes and in various types of casings, and therefore a detection system should be either insensitive to the geometrical shape of the mine and the type of casing material, or preferably provide imaging information. This latter feature will enable the system to better distinguish mines from background clutter, such as rocks, metal shreds, etc. This, in turn, will reduce the false-positive alarm rate and the time wasted in trying to clear an innocuous object thought to be a mine. On the other hand, it is vital that the detection system find a genuine mine; that is, near zero false-negative alarms need to be achieved. Since mines can be buried at different depths under the ground surface, the detection system should not be overly sensitive to the depth of burial. The operator of a detection system should be able to avoid close proximity to the position of the mine to minimize the possibility of inadvertent triggering of the mine. Detection should also be performed at a reasonable operational speed, and at not too prohibitive a cost. In summary, the ideal system must be accurate, not too slow and not too expensive.
     In addition to the detection requirements, practical considerations dictate that a detection system should be easily deployable in the field and be usable by technically unsophisticated people. In other words, the system should not represent a logistical burden by requiring complex machines and operation.
     In summary, mine detection involves dealing with wide variety of mine material and shapes, different soil types and terrain, and non-uniformity of clutter. It is expected that the characteristic signature for the presence of a mine may vary widely depending on local circumstances. It may, therefore, be difficult to apply any one technique unless the nature of the mine, soil and background clutter is well known. It also is desirable to have a technique that is specific in its identification of landmines, so that it is not affected by the surrounding conditions. It is inconceivable, however, that a single detection technology will be able to meet all needs.

A landmine is a foreign object in an otherwise benign medium, the ground. Therefore, the first natural step in mine detection is to sense the presence of an anomaly on or near the surface of the ground, by detecting the existence of an unexpected object. However, this alone is not sufficient to provide a definite indication of mine, as depending on the detection method the identified anomaly could be some other innocuous object. Removing an identified anomaly, with all the care and attention given to a landmine, to discover in vain that the effort was directed towards clearing a harmless object, is a time consuming and costly process (estimates range upwards of $1000 USD per mine cleared). The major problem in demining is to discriminate between a dummy object and a landmine (US Department of State, 1998). Therefore, it is important to characterize a detected anomaly as containing an explosive or non-explosive material. Most common and emerging techniques for landmine detection focus on the detection of anomalies in the ground, without providing a definite indication that an explosive material is present. It is in material characterization that radiation-based techniques can play a vital role in mine detection (IAEA, 1998).
     In the next two sections, common and emerging techniques are summarized and their capabilities are assessed in terms of the two desirable criteria for landmine detection: identification and characterization. Following that, the role radiation-based techniques can play in the mine detection problem is discussed, with regards to the nature of the detection problem and the physical properties of the radiation and material involved.

Currently, four common yet primitive mine detection techniques are widely used: (i) simple visual inspection, (ii) hand-held metal detectors, and (iii) classical mine prodders and (iv) biological sniffers. The limitation of each of these methods is discussed below.
     In visual inspection, one looks for disturbed dirt, apparent mines, etc. Although this is an anomaly identification method, familiarity with the shape and look of a class of landmines can help in characterizing the observed anomaly as likely to contain an explosive charge. The limitations of superficial visual inspection are obvious.
     Metal detectors attempt to obtain information on buried mines by emitting into the soil a time-varying magnetic field to induce an eddy current in metallic objects; which in turn generates a detectable magnetic field. However, landmines typically contain a small amount of metal in the firing pin while many others contain no metal at all. Increasing the sensitivity of a metal detector to detect a smaller amount of metal makes it also very susceptible to metal shreds that are often found in mine infected areas. Although the use of pulsed waveform and monitoring of multi-frequency emissions may improve the capabilities of electromagnetic induction probes, they will remain unsuited for use in magnetic and heavily mineralized soil. Metal detectors, even when successful, can only succeed in identifying the presence of an anomaly, without providing information on whether explosive material is present or not.
     Mine prodders enable subsurface inspection by employing bayonets or hand-held probes (about 250 mm long) to poke the ground, inch by inch, to sense the presence of a hard (solid) object in the soil. It is, therefore, another anomaly identification technique that provides no material characterization information. Prodding is done at an angle to avoid causing detonation if the mine is pushed from the top, where the primary trigger usually is. Aside from the inability of this primitive method distinguish between a landmine and any other solid object that can be present in the soil, such as a rock, it is a dangerous operation.
     Biological sniffing by dogs is also used. Dogs have greater olfactory senses compared to humans, especially for trace quantities, and can be trained to detect the presence of explosives. This is, in effect, a material characterization process as dogs are sniffing the vapours emitted from the explosive material. This technique requires, however, extensive training, and the dogs’ limited attention span makes it difficult to maintain continuous operation. Electronic chemical sniffers can also be used, though they are not as sophisticated as dogs in terms of their detection abilities. Moreover, minefields are usually saturated with residual vapour emissions from recently detonated explosives, which may add to the chemical clutter of the area, thereby confusing the dogs’ senses.
     Overall the detection effort by any of the above widely-used means is tedious and requires disciplined, well-trained personnel. This has motivated international efforts to develop remote detection and turn-key techniques, some of which are discussed in the following section.

A number of mine detection techniques are emerging as alternatives to the currently used methods that have not appreciably changed since World War II.
     The difference in the thermal capacitance between soil and mine affects their heating/cooling rates and therefore their associated infrared emissions. Infrared cameras are used to map heat leakage patterns from the ground which, nevertheless, makes this thermography method an anomaly identification technique (Ashley, 1996). The technique essentially measures the thermal emissivity of the ground and interprets changes in emissivity as being caused by the presence of a foreign object; therefore, material characterization information is not provided. However, this technology has the advantages of being passive, can be performed remotely, by aerial search, and can cover a large area in a short time. Infrared thermography is best suited for identifying minefields (global area search), rather than searching for individual mines (local area search). It cannot however work when the soil and mine are in thermal equilibrium, and therefore is generally limited for use either at sunset or sunrise where a temperature gradient can be established at the ground surface.
     The difference in the reflectance and polarization of soil when disturbed by laser energy may be used to identify the presence of an anomaly (Ashley, 1996). This requires a powerful laser, complex data interpretation and provides no material characterization information.
     Since eddy current can be generated only in conducting materials such as metals and microwaves are completely reflected off metallic surfaces, metal encased land mines can be detected by both pulse-induction metallic detectors and microwaves (ground penetrating radar). Unfortunately, however, not all mines are metallic. Nevertheless, microwaves are also scattered, though to a lesser extent, by non-metallic objects and characteristic refection signatures can be related to material type, and hence can be used to identify explosives. This approach has significant difficulties because of the propagation losses in the soil, the low contrast between target and soil, and the large variety of echoes from the rough surface and other shallow contrasts such as rocks, tree roots, etc. (Peters et. al., 1994). The discrimination of mine from clutter under a wide variety of surface and soil conditions remains very difficult (Peters et. al., 1994). In addition, water has a high affinity to absorbing microwaves, making it difficult to operate ground-penetrating radar under wet conditions. This is an anomaly identification method, with no material characterization ability.
     Although the above methods do not provide material characterization information, experience and familiarity with these methods may enable the reduction of the false-positive detection rate arising from innocuous objects. However, one mistake in such assessment can be fatal and at the end one would have to deal with each identifying anomaly as being a landmine and remove as such; which is a tedious endeavor. It is therefore desirable to have a material characterization method that can determine whether the identified anomaly contains an explosive material or not. Radiation based techniques can be useful in this regard as discussed below.

Penetrating radiation (neutron and photon) offers some attractive features that can be utilized in landmine detection, particularly for material characterization. However, unlike conventional radiographic or tomographic methods, one cannot rely on the radiation transmission modality, as it requires access to two opposing sides of an object; a situation not attainable with landmines. Therefore, one has to rely on secondary radiation emissions (activation) or radiation scattering. The main reason penetrating radiation would be used in landmine detection, in spite of its radiological shielding requirements, is to provide material characterization information. It is therefore useful to closely scrutinize the composition of explosive materials.
     The most common explosive material found in landmines is most likely TNT; although RDX and other plasticized explosives are also used (Ashley, 1996). These explosives are rich in nitrogen, which serves as a bonding agent. However, the amount of nitrogen alone is not sufficient to definitely distinguish an explosive material from other innocuous materials (Hussein, 1992). Explosives are also rich in oxygen (the oxidizing agent). Therefore, knowing the nitrogen content together with the oxygen content provides a more unambiguous identifier of an explosive material (Hussein, 1992). Hydrogen and carbon are also present in most explosives and their relative elemental content may be also used to characterize a detected anomaly as likely to contain an explosive material. The detection of hydrogen and carbon alone is not, however, as a definite indicator of an explosive material as that of the detection of nitrogen and oxygen.
     The composition of soil in which a landmine is embedded varies from dry sand to wet fertile soil. However, the Earth’s crust consists mainly of eight basic elements: oxygen (49.52%), silicon (25.75%), aluminum (7.51%), iron (4.7%), calcium (3.39%), sodium (2.64%), potassium (2.40%) and magnesium (1.94%), while all other elements constitute about 2.15% by weight (Christopher, 1981). Therefore, the only element in the Earth’s crust that also exists in an explosive material is oxygen. However, other objects such as vegetation (e.g. tree roots) and plastic scrap may be also buried in soil. These and other hydrocarbon materials contain elements similar to those found in explosives, and may confuse a material characterization method.
     Some of the ambiguities in composition indication can be reduced by taking advantage of the fact that explosives have a density that is higher than that of most common organic materials, but less than that of metals and many types of soil (Hussein, 1992). Therefore, combing elemental composition information with material density information can be useful in providing definite characterization of an anomaly as containing an explosive material. With these in mind, the use of photons and neutrons in detection is assessed below, with the view that the most effective use of radiation techniques is in determining whether an identified anomaly contains an explosive material or not. Other techniques, discussed above, are more suited for identifying anomalies.

For photons (x- or gamma-rays) to be utilized in landmine detection, they must be used in the scattering or secondary emission modalities. Photons can be emitted as a result of neutron activation, which is discussed in the following section. However, absorption of incident photons can lead to the generation of characteristic x-rays (photopeaks) due to re-arrangement of electrons in the shells of the affected atom. These x-rays are usually re-absorbed within the medium. Bremsstrahlung radiation can also be seen as a secondary emission, generated by the recoil electrons in photon-electron interactions. This process is mainly important in materials with high atomic number, a few of which are abundant in soil, and none of which are present in explosive materials. The third possible photon reaction that leads to secondary emission is the pair production process, in which the electron and positron pair produced annihilate each other generating two 511 keV photons. This reaction is only dominant at high photon energies (greater than about 4 MeV) and for high atomic number materials. Disadvantages of this latter reaction are that it requires the presence of heavy elements which are not abundant in soil and landmines, and high energy photons are required from expensive accelerators that are not easily mountable in field conditions and require special skill and care to operate. In summary, no significant emissions can be expected from photon interactions with matter, and even when such emissions occur they are likely the result of interactions with high atomic-number elements, which are not found in explosive materials.
     The second approach to employing photons in the landmine detection process, excluding photon transmission, is photon scattering. Photon scattering can occur either coherently or incoherently with the electrons of the matter. Coherent (Rayleigh) scattering is a small angle scattering (i.e. forward biased) and would therefore, like transmission, require access to both sides of an interrogated object. It is therefore not a reaction one would anticipate to use for the one-side inspection problem of landmine detection. On the other hand, Compton scattering is dominant in most materials and at almost all photon energies. In this process, photons incoherently collide (Compton scatter) with the atomic electrons with a probability that is dependent on the electron density, and consequently the mass density, of the medium. Therefore, Compton scattering can provide a density map, which can be used as an indicator that an anomaly has a density in the range of explosive materials. This is the essence of the x-ray backscattering system of Campbell and Jacobs (1992). Gamma-rays can also provide similar information, as was shown by Roder (1975).
     As the scattered photons travel back towards the detector they are removed by further scattering or absorption, with the photo-absorption probability being strongly dependent on the atomic number. This approach can provide atomic number (or more precisely effective atomic number) information that can be used for further characterizing a material as being an explosive material. The unique elemental composition of explosives gives them a certain range of effective atomic number. A combination of density and atomic number can then be used to characterize the presence of explosives. This is similar to the approach utilized in commercially available dual energy transmission x-rays systems. The technique is used in some airport inspection and litho-density oil well logging systems (Serra, 1984), and has been proposed for luggage inspection (Hussein, 1994). Localization of the detected object may prove, however, to be difficult when applying this approach, since photons need to propagate over some distance so that the photon energy is sufficiently lowered to the range in which the photoelectric effect becomes dominant (for obtaining atomic number information). Alternatively, a dual energy photon source may be employed, where one energy is higher so that photoelectric effect is not dominant, while the other energy is lower to allow for some photo-absorption while permitting the photons to Compton scatter back into the detector.
     While the above approaches are worth pursuing further, it should be kept in mind that photons will provide mainly density and atomic number information; but no element-specific characterization. This may be sufficient for landmine detection if used in examining an already identified anomaly. For more information on these possibilities, advantages and limitations of photon techniques, the reader can consult the review paper by Hussein and Waller (1998).

The advantage of neutrons over photons is that they can provide elemental information. This information can be used for detecting explosives through the characteristic photon emissions produced by neutron activation. Since explosives used in landmines are rich in nitrogen content, activation of nitrogen by neutron capture can be used for its detection. Absorption of low-energy (thermal) neutrons results in the emission of 10.8 MeV gamma rays, which are easily distinguishable. This is the basis of the thermal neutron activation analysis system, termed "MineSCANS", developed by SAIC Canada (Waller, 1998). The portable isotopic neutron spectroscopy system, named "PINS" (Caffrey et. al., 1992), developed by the Idaho National Engineering & Environmental Laboratory (INEEL), which employs a ²⁵²Cf neutron source (whose neutrons are moderated by the interrogated object) for assaying chemical weapons, may be also tailored for use in landmine detection.
     Thermal neutrons can also be used to activate hydrogen and silicon (producing 2.22 and 3.54 MeV dominant photons, respectively). All military explosives contain from 2 to 3% (by weight) hydrogen, while the ground may contain various concentrations of hydrogen, depending on its water content. Silicon, however, is absent in explosive materials and is present only in soil. Therefore, a thermal neutron activation system may be used to monitor the hydrogen-silicon ratio to search for an anomaly (IAEA, 1998). However, since anomalies can be identified by other non-nuclear means, such an approach will not be helpful in characterizing an anomaly as a landmine. For example, a piece of wood may give the same hydrogen content, and even nitrogen content, as a landmine, with no indication of the presence of silicon. It should be noted that the other two elements present in explosive materials, aside from hydrogen and nitrogen, i.e. carbon and oxygen, have too low an activation (photon-production) cross section at the thermal neutron energy to make them useful in elemental identification with thermal neutrons.
     Thermal neutron activation requires the employment of a bulky moderating material to slow down fast neutrons emitted from either an isotopic source or neutron generator, since there are no portable means of directly generating thermal neutrons. The soil itself can be used as a moderating material, but then the amount of activation will depend on the type of soil (in particular, its hydrogen content). Since the activation probability (cross section) is relatively low, a strong neutron source is required. This causes some difficulties in radiological shielding and handling which affects the portability of the device. Moreover, nitrogen is present in fertile soil and tree roots. Under such conditions it becomes difficult to detect mines based on nitrogen content alone. Moreover, as mentioned earlier, nitrogen and oxygen are present in vegetation, such as tree roots, and therefore the detection of their presence may not be sufficient to characterize an explosive material.
     Fast neutron activation, mainly through the inelastic scattering of neutrons, can be used to provide characteristic gamma rays to identify other elements present in explosive materials. The cross sections of two of the major elements in explosives, nitrogen and oxygen, have characteristic resonances in the energy range from 1 to 3 MeV. Nitrogen has resonance peaks at 1.116, 1.184, 1.593, 1.783 MeV, while oxygen has peaks at 1.312, 1.651, 1.832, 1.907 MeV and a dip at 2.360 MeV (Gomberg and Kushner, 1991). The use of fast neutrons to provide elemental composition is not a new concept (Hussein, 1992) and fast neutron activation has been suggested for elemental identification of explosives (Sawa and Gozani, 1991). However, the activation cross section (interaction probability) of fast neutrons is low for the elements of interest, thus necessitating a high intensity neutron source.
     The PELAN (Pulsed Elemental Analysis with Neutrons) technique (Vouropoulos et. al., 1997) uses fast neutron activation to detect the presence of hydrogen, carbon, nitrogen and oxygen. The technique utilizes a 14 MeV neutron pulse from a portable generator to activate the carbon and oxygen present in the material, since the activation cross section at this neutron energy is relatively high. When the pulse is stopped, the fast neutrons will continue to diffuse within the interrogated medium and its surroundings, and depending on the size and nature of the material encountered, may be slowed to a sufficiently low neutron energy (below 1 eV) where they can be readily captured by hydrogen and nitrogen. The gamma-rays emitted during and after the neutron pulse can be used to detect the four basic elements of explosives provided the soil has sufficient hydrogen content to slow the neutrons down. However, as the pulsed neutrons diffuse from the soil, the number of slow neutrons bombarding an interrogated anomaly may be too low to allow for a definite post-pulse activation analysis.
     The neutron scattering cross section is generally higher than that of activation. Moreover, the backscattering cross section is in general higher than the average scattering probability, at all angles (Gomberg and Kushner, 1991). This makes resonance backscattering an attractive candidate for a single-side access application such as mine detection. By measuring the energy spectrum of backscattered neutrons and monitoring the number of neutrons appearing at the above resonance energies, one can determine whether a large concentration of both nitrogen and oxygen is present in the soil. A number of detectors, surrounding a single source, can then be used to provide an image of the interrogated area. Gomberg and Kushner (1991) took advantage of these resonances in the scattering cross section to distinguish explosive from other innocuous materials in luggage. A similar approach has been used by Brooks et al. (1998) for the detection of explosives in small (0.2-1 kg) samples. Time-of-flight measurements were used by these workers to determine the neutron energy, and required an accelerator-based neutron source. Such sophisticated techniques are difficult to apply in the field conditions of mine detection. However, the use of backscattering for material identification is particularly suited for mine detection, and further effort should be made to develop less complex systems that take advantage of backscattering resonances without the need to perform time-of-flight measurements, thus enabling the use of common radioisotopic sources and conventional neutron detectors.
     Total removal (absorption and scattering) of neutrons offers the advantage of incorporating all interaction modalities, and hence can provide a stronger signal than activation and angle/energy-dependent scattering, for the same source strength. Neutron removal is the process employed in transmission techniques. Although transmission is impossible to apply in the one-side inspection problem of landmines, the concept of measuring the overall effect on an incident signal should also be explored. For example, it may be possible to observe the effect of the resonances in the cross sections of three of the four main elements of explosives (C,N,O) in the energy range of 1 to 3 MeV from energy spectra of isotopic sources, such as ²⁵²Cf and ²⁴¹Am/Be. Cf-252 has an average energy of about 2.14 MeV and spans an energy range from about 0.1 to 6 MeV. For ²⁴¹Am/Be, the average energy is about 4.5 MeV and its energy spectrum spans a range from 2 to 10 MeV. By measuring the energy spectrum of backscattered neutrons from one of these sources, and observing perturbations in the neutron spectrum due to the scattering resonances in the cross sections of the elements of interest, one may be able to produce a "fingerprint" characteristic signature of an explosive. This is similar to the neutron transmission technique used by Gokhale and Hussein (1997) and is currently being investigated by the authors.
     One other detection approach is to simply detect the hydrogen content in soil by neutron slowing down. Although, hydrogen is not a unique indicator of explosives, use can be made of the fact that the level of hydrogen content in soil, landmine and the environment (moisture, vegetation, tree roots, etc.) is quite different. Brooks (1998) measured the change in hydrogen in soil by measuring the intensity of low energy neutrons reflected back from soil exposed to ²⁵²Cf neutrons. They plan to design and test a low cost, hand held device based on this approach.
     Among the radiation interrogation approaches, neutrons are the most suitable candidates for providing composition information non-intrusively, as they directly interact with the nucleus. Neutron-based devices tend, however, to be bulky due to radiological shielding or in some cases moderation requirements. They also tend to be relatively expensive. However, given their potential, efforts should continue to develop compact and inexpensive devices. For information on the potential application of neutron methods, see the paper by Hussein and Waller (1998).

The challenge for radiation physicists is not only to develop techniques that can meet the demanding detection problem, but also to tailor such techniques to their local conditions. After all, detecting landmines in the sandy desert of Egypt or Kuwait is very different from finding them in the fertile soil of Vietnam or Laos. In general, one can state the following:

• Radiation-based techniques have the unique ability to identify the presence of explosives in anomalies detected by other methods.

• It will be difficult to apply any one-detection technique unless the nature of the mine, soil and background clutter is well known.

• Deeply buried landmines may cause future problems, as they are difficult to detect.

• Special "proofing" detection technology may be needed, to look for landmines that may have escaped the demining process.

• It is inconceivable that a single detection technology will be able to meet all needs.

• A fusion of systems is needed: large area scan (remote aerial), smaller area (remote ground-vehicle), and then local scans (man-portable). For an unmanned aerial vehicle, infrared thermography, or laser sensors may be suited, while for ground vehicles, forward looking infrared (FLIR), ground penetrating radar, or pulsed-induction metal detectors can be employed. Man-portable machines are likely to involve metal detectors, FLIR, and ground-penetrating radar. Standoff sensors are likely to include x-ray back scattering, laser acoustics, neutron activation and or scattering, and artificial sniffing biosensors.

• In developing a system that has a reasonable chance for being implemented, the system needs to be cost effective, easy to operate and maintain, dependable, and provide results within a reasonable time.

Ashley, S. (1996) Searching for land mines, Mechanical Engineering, Vol. 118, No. 4, 62.

Brooks, F.D. (1998) Hydrogen Detectors for Landmines, personal communication.

Brooks, F.D., Buffler, A., Allie, M.S., Bharuthram, K., Nchodu, M.R. and Simpson, B.R.S. (1998) Determination of HCNO Concentrations by Fast-neutron Scattering Analysis, Nuclear Instruments & Methods in Physics Research, Section A, Vol. 410, 319-328.

Caffrey, A.G., Cole, J.D., Gehrke, R.J. and Greenwood, R.C. (1992) Chemical Agent and High Explosive Identification by Spectroscopy of Neutron-Induced Gamma Rays, IEEE Transactions on Nuclear Science, Vol. 39, pp. 1422-1426: Caffrey, A.J. et al. (1994) US Army Experience with the PINS Chemical Assay System, INEL report EGG-NPR-11443.

Campbell, J.C. and Jacobs, A.M. (1992) Detection of Buried Land Mines by Backscatter Imaging, Nuclear Science and Engineering, Vol. 110, 417.

Christopher, C.M. (1981) Engineering Geology, Merrill Publ., Columbus, Ohio.

Gokhale, P. and Hussein, E.M.A. (1997) A Californium- 252 Neutron Transmission Technique for Detection of Explosives, Applied Radiation and Isotopes, Vol. 48, 973-979.

Gomberg, H. and Kushner, B. (1991) Neutron elastic scatter (NES) for explosive detection systems (EDS), Proc. First Int. Symposium on Explosive Detection Technology, S. Khan, Ed., DOT/FAA/CT-92/11, FAA Technical Center, Atlantic City, NJ, 123.

Hussein, E.M.A. (1994) Inspection of Luggage using Gamma Radiation", Proc. Cargo Inspection Technologies, Lawrence, A.H., Ed., The International Society for Optical Engineering, San Diego, California, Vol. 2276, 321-325.

Hussein, E.M.A. and Waller, E.J. (1998) Review of One-Side Approaches to Radiographic Imaging for Detection of Explosives and Narcotics, Radiation Measurements, Vol. 26, 581-591.

Hussein, E.M.A., (1992) Detection of explosive materials using nuclear radiation: a critical review", Proc. Conf. on The Aviation Security Problem and Related Technologies, W.H. Makky, Ed., SPIE, Vol. CR42, 126.

IAEA (1998) Detection of Explosives (in particular Landmines) by Low Cost Methods, IAEA/PS/AGM97-3, International Atomic Energy Agency, Vienna, Austria.

King, C. (1998) Demining: Enhancing the Process, IEE Second International Conference on the Detection of Abandoned Landmines, IEE Conference Publication No. 458, Edinburgh, UK, 1-9.

MineFacts (1995) MineFacts CD-Rom Version 1.2, National Ground Intelligence Center, Charlottesville, VA.

Peters, Jr., L., Daniels, J.J. and Young, J.D. (1994) Ground Penetrating Radar as a Subsurface Environmental Sensing Tool, Proc. of the IEEE, Vol. 82, 1802.

Roder, F.L. (1975) Theory and Application of X-ray and Gamma-ray Backscatter to Landmine Detection, Report No. 2134, U.A. Army Mobility Equipment Research and Development Center, Fort Belvoir, Virginia.

Sawa, Z.P. and Gozani, T. (1991) PFNA technique for the detection of explosives, Proc. First Int. Symposium on Explosive Detection Technology, S. Khan, Ed., DOT/FAA/CT-92/11, FAA Technical Center, Atlantic City, NJ, 83.

Serra, O. (1984) Fundamentals of Well-log Interpretation, 1. The Acquisition of Logging Data, Chapter 12, Elsevier, Amsterdam.

US Department of State (1998) Hidden Killer, the Global Land Mine Crisis, U.S. Department of State Publication 10575.

Vourvopoulos, G., Dep, L., Paschal, J. and Spichiger, G. (1997) PELAN- A Transportable, Neutron-Based UXO Identification Technique, Proceedings of UXO Forum'97, Nashville, TN, 342-349.

Waller, E.J. (1998) MineSCANS – Mine System Countermeasures by Analysis of Nitrogen Signature, presented at the 1998 International Advanced Studies Institute Science and Technology Series, Monterey, CA.