LONG RANGE DETECTION AND IDENTIFICATION OF UNDERWATER MINES USING VERY LOW FREQUENCIES (1-1 khz) Timothy J. Yoder' Joseph. A. Bucaro', Brian H. Houstonb, and Harry J. Simpsonb a SFA Inc., Largo, MD; b Naval Research Laboratory, Washington, DC ABSTRACT The Naval Research Laboratory is using its world-renowned structural acoustics facilities (originally developed for scaled submarine programs) to study the broad band (1-15 khz) acoustic scattering from proud and buried underwater mines. The objective is to discover what information is contained in the broad-band properties of the scattered signal which might be exploited for target identification purposes. Current acoustic mine-hunting systems form acoustic images that replicate the rough geometric shape of the target. To obtain sufficient resolution, these systems must operate at frequencies that are too high for anything but time-consuming, close-in looks at the target. Even then, they often confuse mines with minelike targets such as oil drums. In contrast, structural acoustic clues such as mine resonances, elastic wave propagation, internal structure scattering, etc., are available at lower frequencies (1-1 khz), allowing for much longer ranges of operation as well as the construction of unique "fmgerprints" by which to identify the target as a mine. Additionally, at lower frequencies the ocean sediment is more readily penetrated by acoustic waves, creating the possibility for buried mine detection. This paper examines the feasibility of exploiting such very low frequency structural acoustic clues for long range identification ofproud and buried mines. Keywords: mine countermeasures, MCM, underwater mines, buried mines, structural acoustics, classification, identification. 1. ADVANTAGES OF LOW FREQUENCY STRUCTURAL ACOUSTIC IDENTiFICATION The structural acoustic approach to target identification differs fundamentally from the various acoustic imaging techniques currently used in mine hunting operations. Traditional acoustic imaging, like optical imaging, uses a two-step process to form an image for identifying a target as a mine: (1) The incident energy is diffracted by the target in a manner determined by the detailed target shape and associated reflectivity; and (2) the diffracted energy is brought to a focus with a lens or an array to form the fmal image. The ability to resolve the object's spatial detail is proportional to the product ofthe illuminating frequency and the size of the lens or array. Thus, traditional imaging systems use high frequencies (with wavelengths much shorter than the desired spatial resolution) in order to limit the size of the focusing element needed to "see" the geometrical detail. The necessity of using high frequencies or a larger focusing element to achieve the resolution required to identify targets limits high resolution imaging systems to very short ranges because either high frequency sound is quickly attenuated in the ocean or an inhibitively large focusing element is needed. a Further author information - T.J.Y.: Email yoder@volley.nrl.navy.mil; Telephone: (22) 767-2492; Fax: (22) 767-2492. J.A.B.: Email jbucaro@ccf.nrl.navy.mil; Telephone: (22) 767-2491; Fax: (22) 767-2492. B.H.H.: Email houston @nrl.navy.mil; Telephone: (22) 44-85 1 ; Fax: (22) 44-855 H.J.S.: Email Hany.Simpson@nr1.navy.mil; Telephone: (22) 44-1541; Fax: (22) 44-855 Part of the SPIE Conference on Detection and Remediation Technologies for Mines and Minelike Targets Ill Orlando, Florida April 1998 SPIE Vol. 3392 277-786X1981$l. 23
In contrast, the structural acoustic approach uses low frequencies (with long wavelengths) to excite an internal response from the target. This idea is best understood from the following example. Consider two soda cans, one full and one empty. It is easy to tell the full can from the empty one by simply tapping the cans and listening to the responses. It is impossible, however, to tell the difference by looking at them, i.e., a normal image is useless. Likewise, the unique internal structure of a mine should produce an acoustic fmgerprint that can be used to distinguish it from a myriad of false targets such as oil drums, rocks, pier pilings, etc. all of which produce acoustic images similar to the mine. While close-in, high resolution images can often be used to separate these targets, it is impossible to differentiate them via acoustic imaging even at moderate ranges, say tens of meters. This ambiguity causes mine clearing operations to be halted for time-consuming and potentially risky close-in looks at every false target and live mine. The structural acoustic approach includes potential advantages beyond lowering the false alarm rate at moderate ranges. Since the structural acoustic approach uses low frequencies, the transmitted and scattered sound is not highly attenuated. This opens the possibility for long range identification and detection to hundreds or even thousands of meters. Long range acoustic identification and detection can be further enhanced because the water column, acting like a wave guide, can be exploited in a number of ways to cause the transmitted sound field to focus on any potential target, either proud or buried. The identification and detection ofburied mines is another advantage of using the structural acoustic approach. Buried mine detection and identification are possible because low frequency acoustic waves more readily penetrate the sediment and because some structural acoustic fmgerprints become enhanced when an object is buried. Contrasted to this, high frequency waves used in acoustic imaging are so highly attenuated by the sediment that they are almost useless for detecting buried mines. 2. IDENTIFICATION VIA STRUCTURAL ACOUSTICS Mine identification via structural acoustic clues requires separation of echoes based on the internal response ofthe target. This identification can be as simple as the auditory process used to distinguish full soda cans from empty ones, or it can be a more sophisticated computer process that automatically identifies targets. We present one preliminary example of structural acoustic identification which differentiates a water-filled oil drum sitting proud on a sandy bottom from a group of mine-like targets (either in the freefield, proud, or halfburied) with complicated internal structures. 2.1 Experimental Facilities The Freefield Facility and the Porous Media Bottom Facility (both which reside in the conglomerate 1 experimental facilities called the Laboratory for Structural Acoustics1 operated by the Physical Acoustic Branch at the Naval Research Laboratory) were used in collecting the scattering data analyzed for structural acoustic identification. The Freefield Facility consists of an indoor cylindrical tank (16.76 meters in diameter and 15.24 meters deep) which is filled with approximately 1 million gallons of deionized water. The facility has 1) vibration and temperature isolation through specially designed mounts which suspend the tank above its foundation, 2) feedback controlled heating elements and adiabatic materials on the walls to reduce temperature variability, and 3) anechoic materials on the walls to reduce reverberation. This configuration provides a very homogenous acoustic medium with a temperature variability of less than one hundredth of a degree over time periods of up to several weeks. This facility is also configured with robotics scanners and target manipulators. A large three dimensional workspace (4.3 meters by 4.3 meters by 2.4 meters) scanner with a spatial repeatability tolerance of 13 microns is employed for nearfield acoustic holography (NAH)2'3 and laser doppler vibrometry (LDV)4 measurements. A bistatic scanner is specifically designed to collect a full 36 degrees of data in a circle around a target with a repeatability tolerance of six ten thousandth of a degree. The target manipulators can rotate a target through a full 36 degrees with the same repeatability tolerance (six ten thousandth of a degree) as the bistatic scanner. This combination of precision computer controlled robotics and painstaking measures to insure homogeneity and stability produce some of the worlds highest fidelity and unique underwater acoustic measurements. 24
The Porous Media Bottom Facility consists of an indoor rectangular tank approximately 1 meters long, 8 meters wide, and 7 meters deep with anechoic walls. The facility has 3. meters of sand on the bottom and 3.8 meters ofdeionized water above the sand. The sand was purchased from a commercial manufacturing company, Unimin, and consists of 3 tons of washed sand produced via a crushing process and sized using sieves. The mean grain diameter of the sand is 212mm which is a typical mean grain diameter for many off shore sands. (A complete description of the bottom characteristics can be found in the following reference5.) This facility has homogeneous water and bottom characteristics with a temperature variability of less than one tenth of a degree over time periods of up to several weeks. The facility is configured with a 4.6 meter by 5.2 meter two-dimensional Cartesian robotics system with a repeatability tolerance of 26 microns. This robotics system is used to position receiver hydrophones in a horizontal plane at the working depth of a measurement. Once again, the combination of precision computer controlled robotics and painstaking measures to insure homogeneity and stability produce high fidelity underwater measurements in a porous media facility. 2.2 Experimental Set-up Bistatic scattering data was collected from a water filled oil drum sitting proud on the sandy bottom. The experimental arrangement is shown in Figure 1. The source insonifies the oil drum end-on, and the receiver is moved around the oil drum starting at the monostatic location (r) through forward scattering (Or=l 8 ) and ending back at the monostatic location (8r36 ). The source and receiver are 3.16 meters and 2. meters, respectively, from the center of the oil drum and remain in the plane that is 25 cm above the sand throughout the measurement. Bistatic scattering data with experimental set-ups similar or identical to the oil drum experiment was collected from a group of mine-like targets (either in the freefield, proud, or half buried) with complicated internal structures. The proud and half buried data was collected in the Porous Media Bottom Facility with identical source and receiver configurations as the oil drum experiment. The freefield data was collected in the Freefield Facility with the experimental arrangement shown in Figure 2. The only differences between the freefield data and the sandy bottom data (other than the sandy bottom) are the source, receiver, and the distance from the target center to the source and receiver which are 3.27 meters and 2.77 meters, respectively. 2.3 Data Analysis A multiple step process is undertaken to obtain the high fidelity bistatic scattering impulse response from a target. First, an impulsive pressure wave which is maximally flat over a wide frequency band is created at the target center's location. This is accomplished by measuring the incident pressure wave from the source at the target center's location when it is driven by a spectrally flat impulsive voltage. An inverse filter technique is then used to generate a new waveform which drives the source to produce an impulsive pressure wave which is maximally flat over a given frequency band. This new pressure wave is measured at the target center's location and stored for later processing. Next, the target is placed in the facility and the total field is measured and coherently averaged one hundred times for each receiver location. (Averaging the signal one hundred times produces a SNR of more than 7 db.) After measuring the total field, the target is removed and the incident field is measured (and averaged) for each receiver location. The scattered field is then simply found by coherently subtracting the incident field from the total field. The SNR level of the scattered field found this way is greater than 55 db. This high fidelity coherent subtraction is only possible because of the painstaking measures taken in our facilities to insure repeatability, stability, and spectrally flat incident pressure waves. Finally, the incident waveform (measured at the target center's location) is deconvolved with the scattered field to produce the impulse response of the target which has greater than a 55 db SNR. The target response is purposely measured in the nearfield using our facilities' "compact range" measurement capability. This nearfield data is used to project the scattering in to the farfield (or any intermediate range desired) using a Huygens-based algorithm6. Simply stated, the algorithm determines the frequency dependent source distribution needed to produce the measured nearfield data. Then the algorithm uses this source distribution to calculate the scattered pressure at any desired location. The facilities "compact range" measurement capability combined with a version of this algorithm results in a capacity to use one target measurement to calculate the scattering from the target in a number of different 25
environments. This is accomplished by modifying the second part of the algorithm so that the source distribution is used to calculate the scattered pressure given the environmental parameters. Figure 3 displays the farfield bistatic target strength ofthe end insonified proud oil drum. For angles near backscattering, there is a strong beating pattern evident whose amplitude decreases with frequency. This beating pattern is caused by the reflection from the top of the oil drum interfering with the reflection from the bottom of the oil drum. A simple submerged infmite plate reflectionltransmission model can be used to explain the pattern. Part of the energy incident on the top of the oil drum scatters back to the receiver and part of the energy is transmitted. The transmitted energy is, in turn, partly reflected by the bottom ci the oil drum and then partly transmitted through the top of the oil drum to the receiver. A first order approximation for the scattered pressure from the two signals is: P R+TRTeC A where P is the scattered pressure, A is the wavelength, co is the angular frequency, c is the speed of sound in water, d is the separation between the top and bottom of the oil drum, A is the area of the top and bottom ofthe oil drum, R is the frequency dependent reflection coefficient, and T is the frequency dependent transmission coefficient. Both the frequency dependence of the amplitude (Rocl/t) and the expected time separation of the two reflections (d/c) can be easily seen when the backscattered data is displayed in the joint time-frequency domain (the lower left mosaic in Figure 4). The joint time-frequency displays in Figure 4 can be thought of as acoustic fmgerprints which display the acoustic color or frequency content as a function of arrival time for an echo. Figure 4 displays a typical acoustic fmgerprint constructed with data from the oil drum and a mine-like target. The fmgerprint from the mine-like target has a large low frequency highlight caused by the complicated internal structure of the target. This highlight is absent from the oil drum fingerprint because the internal structure is different. The existence of the internal structure highlight is used in a simple algorithm to automatically differentiate between the mine-like targets and the oil drum. Specifically, the ratio of the acoustic color at two frequencies is plotted as a function oftarget angle for a group ofmine-like targets and the oil drum in Figure 4. The simple threshold value displayed in the figure clearly separates these two groups of targets and demonstrates structural acoustic identification of a group of mine-like targets operating against what is considered to be a difficult false target. 3. CONCLUSIONS This preliminary example of low frequency structural acoustic identification which differentiates a waterfilled oil drum sitting proud on a sandy bottom from a group of mine-like targets (either in the freefield, proud, or half buried) with complicated internal structures demonstrates that freefield, proud, and half buried underwater targets can be acoustically identified without the need for acoustic imaging. This low frequency approach opens the possibility for long range identification and detection to hundreds or even thousands of meters because the transmitted and scattered sound is not highly attenuated and because an inhibitively large focusing element is not needed. The pay off from a structural acoustic approach to mine counter measures is due to the long range identification potential of both proud and buried mines. The new ability to detect and identify buried mines can reduce serious risk to the fleet. In addition, long range identification would greatly increase the speed of mine clearing operations and reduce risk to assets by eliminating the closerange operations currently used to identify a target. This work is supported, in part, by ONR Code 321. 4. ACKNOWLEDGMENTS 26
5..REFERENCES 1. B. H. Houston, "Structural acoustic laboratories at NRL in Washington, D.C.", J. Acous. Soc. Am. 92(4), October 1992. 2. E. G. Williams, B. H. Houston, and J. A. Bucaro, "Broadband nearfield acoustic holography for vibrating cylinders", J. Acous. Soc. Am. 86(2), August 1989. 3. D. M. Photiadis, J. A. Bucaro, and B. H. Houston, "Scattering from flexural waves on a ribbed cylinder", J. Acoust. Soc. Am. 96(5), November. 1994. 4. J. F. Vignola and B. H. Houston, "The Design of a Three Dimensional Laser Vibrometer", Accepted August 1993 ASME Journal of Vibration and Acoustics. 5. H. J. Simpson and B. H. Houston, "Synthetic array measurements of acoustic waves propagating into a water-saturated sandy bottom for a smoothed and a roughened interface", J. Acoust. Soc. Am. in review. 6. E. G. Williams, L. Kraus, and B. H. Houston, "Radix Range Correction Algorithm applied to SARA2D Target Strength Data", Naval Research Laboratory Internal Memo Report, December 1994 27
A Top View F33 Source Br Receiver Side View ( 3.84 M WATER 4' BNK 813 Receiver A F33 Source V - - /.25 M ;.25 M 3. M SAND Figure 1: Porous Media Bottom Facility Experimental Set-up 4 9M NF13 Plane Wave Source 3.27 M Or. F68 Receiver 'V Figure 2: Freefleld Facility Experimental Set-up 28
1 5 1 NI >-, C-) ci.) a) LL Bistatic Scattering Angle (Degrees) Figure 3: Bistatic Target Strength of a Proud Water-Filled Oil Drum 29
Mine-Like Target with internal Structure 1 Identification Space I- (.) 4- U) icz False Target (Oil Drum) without Internal Structure F.1 C.) 4- U) C.) 7 Mine-Like Targets with Inte1nt - Structure / /i \,' / - - Oil Drum Bistatic Angle 5 - Unique Structural Correct Acoustic Identification of "Finger Prints" Targets Figure 4: The structural acoustic "fingerprints" are time-frequency realizations of target echoes. Visual inspection of the acoustic fingerprints clearly shows differences between a complicated mine-like target with internal structure and an oil drum. A simple identification algorithm compares the acoustic color of the echoes and produces a clear separation of the oil drum from five different mine-like targets for multiple bistatic scattering angles. 2 11)