Enhancing dielectric contrat between land mine and the oil environment by watering: modeling, deign, and experimental reult. Brian Borcher a, Jan M. H. Hendrickx b, Bhabani S. Da c, and Sung-Ho Hong b a Department of Mathematic, New Mexico Tech, Socorro, NM 87801 1 b Department of Earth and Environmental Science, New Mexico Tech, Socorro, NM 87801 c Agricultural and Food Engineering Department, Indian Intitute of Technology, Kharagpur, Wet Bengal, PIN 7130, India ABSTRACT The complex dielectric contant of the oil urrounding a landmine and it contrat with the dielectric contant of the landmine are critical to the effectivene of ground penetrating radar (GPR) for landmine detection. Thee parameter affect the velocity and attenuation of the radar ignal a well a the trength of the reflection from the mine. The dielectric propertie of the oil depend on the oil texture and bulk denity a well a the oil water content. In previou work, we have imulated the unaturated water flow around a landmine. In thi paper we ummarize a collection of model that can be ued to predict the dielectric contant, velocity of the GPR ignal, attenuation, and reflection coefficient from oil type and oil water content. Thee model have been integrated into a MATLAB oftware package. Uing thee model, we can determine whether or not field condition are appropriate for ue of GPR. Under dry condition, the oil water content may be too low for good GPR performance. If the oil i too dry, we can elect an appropriate level of oil water content and deign a watering cheme to bring the oil water content up to the deired level. We preent a cae tudy in which a oil watering cheme wa deigned, imulated, and then performed at a field ite. Keyword: Landmine Detection, Ground Penetrating Radar, Dielectric Propertie, Modeling 1. INTRODUCTION The dielectric propertie of the oil urrounding landmine have an important effect on the performance of ground penetrating radar (GPR) ytem for landmine detection. The dielectric propertie of the oil depend on the oil texture, bulk denity, and water content. Although oil texture and bulk denity cannot be changed, it i poible to manipulate oil water content by watering the oil. In ection of thi paper we dicu a uite of model that can be ued to predict the dielectric propertie of oil and the reulting repone of a ground penetrating radar ytem. The model output can be ued to determine whether or not GPR i likely to be an effective enor under particular field condition. In ituation where the oil i too dry for GPR to work well, we can ue the model to determine a water content level that will permit the effective ue of GPR. Given a deired oil water content, we can deign a watering plan that will bring the oil water content up to the deired level. We can alo ue imulation model of unaturated water flow around the landmine to verify that uch a plan i likely to have the deired effect. In ection 3, we preent an example in which thi approach wa ued to deign a watering plan for a dry and oil at the Sevilleta National Wildlife Refuge. The watering cheme wa imulated to verify that it would bring oil water content up to the deired level. We then performed a field experiment in which the watering plan wa followed and direct meaurement of oil water content were obtained. The experimental reult confirmed the model prediction. 1 Correpondence: B. Borcher. Other author information: B. B.; Email: borcher@nmt.edu; Telephone: 505-835- 5813; Fax: 505-835-5366. J. M. H. H.; Email: hendrick@nmt.edu. S. H. H.; Email: hong@nmt.edu. B. S. D; Email: bda@agfe.iitkgp.ernet.in. 1
. MODELS OF SOIL DIELECTRIC PROPERTIES AND RADAR RESPONSE The dielectric propertie of a oil depend on a number of factor, including it bulk denity, the texture of the oil particle (and, ilt, or clay), the denity of the oil particle (typically about.6 g/cm 3 ), the volumetric water content of the oil, the temperature, and the frequencie of interet. 1-3 Recent reearch ha alo hown that the dielectric propertie of oil depend on the amount of bound water which i in cloe contact with mineral in the oil. 4-5 Theoretical and empirical model of the dielectric propertie of the different component of the oil have been combined into emiempirical mixing model which can be ued to predict the dielectric propertie of field oil. 1,4,5,6,7 In thi ection we ummarize the 1995 model of Peplinki, Ulaby, and Dobon. 7 Thi model i a variant of the earlier model of Dobon, Ulaby, Hallikainen and El-Raye. 5 The earlier model wa calibrated for frequencie in the range of 1.4 to 18 Ghz. 8 The model dicued here wa calibrated by fitting the model to a et of experimental obervation with a variety of oil texture, oil water content, and frequencie from 0.3 to 1.3 Ghz. 7 The input to thi model conit of the volumetric water content, the frequency f, the fraction of and particle S, the fraction of clay particle C, the denity of the oil particle S (a typical value i.66 g/cm 3 ), and the bulk denity of the oil B. An empirically derived formula for effective conductivity i eff 0.04670.04 B 0.4111S0.6614C. (1) The and and clay fraction alo enter the model through two contant which depend on the oil type but are independent of the frequency and oil water content. 17.480.519S0.15C () 1.337970.603S0.166C (3) The real and imaginary part of the dielectric contant for the free water are given by where fw fw fw i (4) fw w w0 w 1( f w ) (5) and fw f w w0 w eff 1( f w ) 0 f ( S B ) S (6) In thee formula, 0 i the dielectric permittivity of free pace, w0 i the tatic dielectric contant of water (80.1 at 0C ), w i the high frequency limit of fw (4.9), and w i the relaxation time of water ( 9.3 10 1 at 0C.) The dielectric contant of the oil particle i given by the empirical model (1.010.44 ) 0.06. (7)
Finally, the real and imaginary part of the dielectric contant for the bulk oil are etimated by i (8) where 1.15 1 B S 1 fw 1/ 0.68 (9) and In thee formula, 0.65i a contant that ha been empirically fitted to the data. 1/ fw. (10) Given the complex dielectric permittivity of the oil and the dielectric contant of a landmine buried in the oil, we can attempt to model the repone of a GPR ytem. The repone of the radar depend in complicated way on tranmitted power, the frequency at which the radar operate, the geometry of the radar tranmitter and receiver antenna, the depth to the mine, cattering from the mine, the urface roughne of the oil, and the enitivity of the receiver. In the following, we preent a implified model that incorporate only attenuation effect and a imple model of reflection of plane wave from a flat urface. 9 More ophiticated model incorporating the pecific feature of particular GPR ytem hould be ued when they are available. A GPR ignal travel through the oil, they are attenuated at a rate determined by the complex dielectric contant of the oil. The round trip attenuation lo in db i given by Attenuation Lo17.3718 d (11) where d i the depth to the object from which the GPR ignal i reflecting, and i given by f c 1 1 (1) Notice that the attenuation increae with frequency. The attenuation alo increae with the dielectric contant a oil water content increae. A econd important factor in the performance of GPR ytem for landmine detection i the trength of the reflection from the landmine. For the implet cae of a plane wave, vertically incident on the top of the mine, the reflection coefficient i given by r m m (13) where i the complex dielectric contant for the oil, and m i the dielectric contant of the mine. The reflection lo in db i given by Reflection Lo 10log r (14) Notice that the reflection coefficient depend on the difference between dielectric contant of the mine and the oil. A thee contant approach each other, the trength of the reflected wave goe to zero, and the mine become inviible. The mathematical model decribed here have been integrated into a MATLAB package that can be ued to predict the performance of ground penetrating radar ytem under field condition. The neceary input data conit of the oil texture (in the form of and and clay fraction), the bulk oil denity, and the volumetric oil water content. 3
3.AN EXAMPLE A an example of our approach, we decribe an experiment performed at the Sevilleta National Wildlife Refuge near Socorro, New Mexico. The oil at thi ite i a dry and, with a typical oil water content of 5%, and a oil texture of 95% and, % ilt and 3% clay. The bulk denity of the oil i approximately 1.6 gram per cubic centimeter. The oil texture and bulk denity were determined uing tandard method. 10,11 We conidered the cae of a landmine with a dielectric contant of 6 buried at a depth of 15 cm. Figure 1 how the attenuation and reflection loe for a radar operating at 1. Ghz. The mallet loe occur in a very dry oil with le than 1% volumetric water content and in moit oil with water content exceeding 15%. So the bet condition for land mine detection uing GPR are extremely dry condition or moit condition. Unfortunately, mot oil have field water content near the oil urface that are neither extremely dry nor moit. Field water content of le than 1% will only occur under very arid climatic condition. Another apect of oil water regime i the large patial variability of field oil water content 1. Even if the average oil water content wa near 1%, ome area in the oil would have higher water content. Thu urvey under dry oil condition are not recommended. Since 0 5 Loe a a Function of Theta Attenuation Reflection Total 10 15 0 Lo in db 5 30 35 40 45 50 0 0.1 0. 0.3 0.4 0.5 0.6 Theta Figure 1: Attenuation and reflection loe for the Sevilleta and. 4
typical water content of urface oil hover between % to 15%, we conclude on the bai of the loe in Figure 1 that natural oil condition form a poor environment for landmine detection uing GPR. However, if the and oil i wetted to water content exceeding 0% condition improve coniderably and become inenitive to local variation in oil water content due to patial variability. Since the practice of irrigation on agricultural land i common in mot part of the world, a wide range of technological olution ha been developed to wet field oil in a cot-effective manner which make oil watering a feaible approach to enhancing GPR repone in mine field. With prevailing water content level of around 5% at our tet ite, there i very little contrat between the dielectric contant of the oil and the dielectric contant of the mine. We would like to raie the oil water content to about 0%. A rough calculation how that ince 15% of 40 cm i 6 cm, the application of 6 cm of water hould be ufficient to raie the average oil water content from 5% to 0% throughout the top 40 cm of oil. However, ince ome water will penetrate below 40 cm, and ince the ditribution of water within the oil will not be completely uniform, we plan to apply a total of 10 cm of water. In deigning a watering plan, it i important to take into account how fat the oil can aborb water. For thi and oil, a watering rate of about 5 cm per hour i reaonable. Figure how a chematic view of the experimental profile, including the landmine imulant and four depth at which we will meaure oil water content ( cm, 1 cm, 30 cm, and 37 cm.) The landmine imulant i buried at a depth of 15 cm, and ha a radiu of 15 cm and a height of 8 cm. Figure 3 how the reult of a HYDRUS-D imulation of our watering cheme. 13 We have previouly decribed the ue of HYDRUS-D to imulate the unaturated flow of groundwater around landmine. 14 HYDRUS-D doe not compute a full three dimenional imulation of water flow for general model. Intead, we ued a feature of HYDRUS-D to compute the three dimenional water flow for a model with radial ymmetry around the axi of the landmine. The oil hydraulic parameter that affect the movement of water in the oil were etimated from the oil texture uing a feature of HYDRUS- D. An initial condition of % water content at the urface, ramping up to 7% water content at 40 cm wa ued to initialize the imulation. The figure how that at the end of the watering period, the oil water content hould be raied to around X 0.0m X 0.1m X 0.30m X 0.37m Figure : Layout of the landmine imulant and location of water content meaurement. 5
0.35 1 0.30 10 Volumetric Water Content 0.5 0.0 0.15 0.10 0.05 0.0m 0.1m 0.30m 0.37m Applied Water 8 6 4 0 Cum. Applied Water [cm] 0.00 0 1 3 4 Elaped time [hour] Figure 3: Hydru-D Simulation of water content above and below the mine. 0% throughout the firt 40 cm of the oil. The oil water content i omewhat lower beneath the mine, becaue it take a fairly long period of time for water to move through the oil and around the obtruction. The imulation alo how that within about five hour, the oil water content will drop back down to around 10%. Next, we performed a field experiment in which the ame watering cheme wa ued at the Sevilleta ite. Water wa applied uing a prinkler ytem. Two 50 gallon tank of water were required. There wa a paue in the watering after the firt 10, 15 hour to refill the tank. The oil water content wa meaured uing time domain reflectometry (TDR). Figure 4 how how the oil water content above and below the mine evolved during the watering proce. A expected, the actual oil water content wa near 0% at all monitoring point by the end of the watering proce. Depite the fact that watering the field did not proceed preciely a planned, the watering cheme did generally work a predicted by the imulation. 4. SUMMARY AND CONCLUSIONS In thi paper we have decribed a collection of model which can be ued to predict the dielectric propertie of field oil given information about oil texture, bulk denity, and water content. Given the dielectric propertie of a oil, it i poible to predict how ground penetrating radar (GPR) ignal will be attenuated by the oil. It i alo poible to predict how the contrat between the dielectric contant of the oil and the dielectric contant of a landmine will affect the trength of the reflected GPR ignal. In ituation where the dielectric propertie of the oil make it unlikely that GPR will be an effective enor for landmine detection, we can ue a watering cheme to add water to the oil and change the dielectric contant. It i poible to deign a watering cheme that will bring about the deired change in the oil water content and dielectric contant. We can ue a imulation of the unaturated flow of water through the oil to verify that the watering cheme will produce the deired change in the oil water content. 6
0.30 1 0.5 10 Volumetric Water Content 0.0 0.15 0.10 0.05 0.0m 0.1m 0.30m 0.37m Applied Water 8 6 4 0 Cum. Applied Water [cm] 0.00 0 1 3 4 Elaped time [hour] Figure 4: Meaured oil water content above and below the mine. A an example of thi approach, we conidered a ituation in which an anti-tank landmine imulant wa buried at a depth of 15 cm in a dry and oil. The model of the dielectric propertie of the oil howed that GPR wa likely to perform poorly under thee circumtance becaue of the lack of trong contrat between the dielectric contant of the oil and the mine. The model of the dielectric propertie alo howed that if we could increae the water content of the oil to approximately 0%, then the dielectric contant of the oil would be favorable to GPR detection of the landmine. We deigned a watering plan to bring the oil water content up to 0%. A imulation of the watering cheme verified that the watering plan hould have the deired effect. In a field experiment, we followed the watering plan and found that a expected, the oil water content wa increaed to about 0% both above and below the mine. ACKNOWLEDGMENTS Thi work i funded by a grant from the Army Reearch Office (Project 38830-EL-LMD). The author would like to thank Dr. Ruell S. Harmon, Senior Program Manager at the Army Reearch Office, for hi valuable advice and upport. REFERENCES 1. P. Hoektra and A. Delaney, Dielectric propertie of oil at UHF and microwave frequencie, J. of Geophyical Re., 79, pp. 1699-1708, 1974.. G. C. Topp, J. L. Davi, and A. P. Annan, Electromagnetic determination of oil water content: meaurement in coaxial tranmiion line, Water Reour. Re., 16, pp. 574-58, 1980. 3. F. T. Ulaby, R. K. Moore, and A. K. Fung, Microwave remote ening : active and paive, volume 3, Artech Houe, Dedham, MA, 1986. 4. J. R. Wang and T. J. Schmugge,. An empirical model for the complex dielectric permittivity of oil a a function of water content, IEEE Tran. Geoci. Remote Sen., 18, pp. 88-95, 1980. 5. M. C. Dobon, F. T. Ulaby, M. T. Hallikainen, and M. A. El-Raye, Microwave dielectric behavior of wet oil - Part II: 7
Dielectric mixing model, IEEE Tran. Geoci. Remote Sen., 3, pp. 35-46, 1985. 6. D. Wobchall, A theory of the complex dielectric permittivity of oil containing water, the emidipere model, IEEE Tran. Geoci. Electron., 15, pp. 49-58, 1977. 7. N. R. Peplinki, F. T. Ulaby, and M. C. Dobon, Dielectric propertie of oil in the 0.3-1.3GHz range, IEEE Tran. Geoci. Remote Sen., 33, pp. 803-807, 1995. 8. M. T. Hallikainen, F. T. Ulaby, M. C. Dobon, and M. A. El-Raye, Microwave dielectric behavior of wet oil - Part I: Empirical model and experimental obervation, IEEE Tran. Geoci. Remote Sen., 3, pp. 5-34, 1985. 9. A. R. Von Hippel,. Dielectric material and application, Wiley, New York, 1954. 10. J. M. H. Hendrickx, Determination of hydraulic oil propertie, in Proce tudie in hilllope hydrology, M. G. Anderon and T.P. Burt ed.,. pp. 43-9, Wiley, New York, 1990. 11. A. Klute ed., Method of Soil Analyi - Part 1 Phyical and Mineralogical Method (Second Edition), Agronomy Monograph No. 9., SSSA, Madion, WI, 1986. 1. J.M.H. Hendrickx, P.J. Wierenga, and M.S. Nah. Variability of oil water tenion and oil water content. Agricultural Water Management, 18, pp. 135-148, 1990. 13. J. Šimnek, M. Šejna, and M. Th. Van Genuchten, HYDRUS-D/MESHGEN-D code for imulating water flow and olute tranport in two-dimenional variably aturated media. IGWMC - TPS 53C, Ver..01, International Ground Water Modeling Center, Colorado School of Mine, CO 80401, USA, 1999.. 14. B. Borcher, J. M. H. Hendrickx, and B. S. Da, Modeling Ditribution of Water and Dielectric Contant Around Landmine in Homogeneou Soil, in Detection and Remediation Technologie for Mine and Minelike Target IV, A. C. Dubey, J. F. Harvey, J. T. Broach and R. E. Dugan ed. Proceeding of SPIE [3710-70], pp. 78-738, SPIE, Bellingham, WA, 1999. 15. I. White and S. J. Zegelin, Electric and dielectric method for monitoring oil water content, in Handbook of vadoe zone characterization and monitoring, L. G. Wilon, L. G. Everett and S. J. Cullen ed., pp 343-386, Lewi Publiher, Boca Raton, FL, 1995. 8