NAVAL POSTGRADUATE SCHOOL Monterey, California THESIS ENVIRONMENTAL VARIABILITY ON ACOUSTIC PREDICTION USING CASS/GRAB. Nick A.

Transcription

1 NAVAL POSTGRADUATE SCHOOL Monterey, California THESIS ENVIRONMENTAL VARIABILITY ON ACOUSTIC PREDICTION USING CASS/GRAB by Nick A. Vares June 2002 Thesis Advisor: Second Reader: Peter C. Chu Steve Haeger (NAVO) Distribution authorized to DoD and DoD Contractors only; Critical Technology June Other requests for this document must be referred to Superintendent, Code 0052, Naval Postgraduate School, Monterey, CA via the Defense Technical Information Center, 8725 John J. Kingman Rd., STE 0944, Ft. Belvoir, VA

2 THIS PAGE INTENTIONALLY LEFT BLANK

3 REPORT DOCUMENTATION PAGE Form Approved OMB No Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA , and to the Office of Management and Budget, Paperwork Reduction Project ( ) Washington DC AGENCY USE ONLY (Leave blank) 2. REPORT DATE June TITLE AND SUBTITLE: Title (Mix case letters) Environmental Variability on Acoustic Prediction Using CASS/GRAB 6. AUTHOR(S) Vares, Nick A. 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Naval Postgraduate School Monterey, CA SPONSORING /MONITORING AGENCY NAME(S) AND ADDRESS(ES) NAVOCEANO Stennis Space Center, MS REPORT TYPE AND DATES COVERED Master s Thesis 5. FUNDING NUMBERS 8. PERFORMING ORGANIZATION REPORT NUMBER 10. SPONSORING/MONITORING AGENCY REPORT NUMBER 11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. 12a. DISTRIBUTION / AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE Distribution authorized to DoD and DoD Contractors only; Critical Technology June Other requests for this document must be referred to Superintendent, Code 0052, Naval Postgraduate School, Monterey, CA via the Defense Technical Information Center, 8725 John J. Kingman Rd., STE 0944, Ft. Belvoir, VA ABSTRACT (maximum 200 words) The goal of this research is to examine the environmental effects on shallow water bottom moored mine detection using the Navy s Comprehensive Acoustic Simulation System/Gaussian Ray Bundle (CASS/GRAB) model for a generic Very High Frequency (VHF) forward looking sonar. The effects of imprecise bottom type and wind speed data are evaluated to determine the impact of this variability on mine detection. The results show that signal excess variability is small and operational benefits may be maximized with a slightly better sonar. The effects of shallow and deep transducer placement in the water column are compared to determine which yields a greater probability of mine detection. The greatest probability of mine detection exists for the deep transducer. 14. SUBJECT TERMS CASS/GRAB, Modeling and Simulation, Oceanography, Mine Warfare 17. SECURITY CLASSIFICATION OF REPORT Unclassified 18. SECURITY CLASSIFICATION OF THIS PAGE Unclassified 19. SECURITY CLASSIFICATION OF ABSTRACT Unclassified 15. NUMBER OF PAGES PRICE CODE 20. LIMITATION OF ABSTRACT NSN Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std UL i

4 THIS PAGE INTENTIONALLY LEFT BLANK ii

5 Distribution authorized to DoD and DoD Contractors only; Critical Technology June Other requests for this document must be referred to Superintendent, Code 0052, Naval Postgraduate School, Monterey, CA via the Defense Technical Information Center, 8725 John J. Kingman Rd., STE 0944, Ft. Belvoir, VA ENVIRONMENTAL VARIABILITY ON ACOUSTIC PREDICTION USING CASS/GRAB Nick A. Vares Lieutenant, United States Navy B.S., University of Memphis, 1994 Submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE IN PHYSICAL OCEANOGRAPHY from the NAVAL POSTGRADUATE SCHOOL June 2002 Author: Nick A. Vares Approved by: Peter Chu, Thesis Advisor Steve Haeger (NAVO), Second Reader Mary Batteen, Chairman Department of Oceanography iii

6 THIS PAGE INTENTIONALLY LEFT BLANK iv

7 ABSTRACT The goal of this research is to examine the environmental effects on shallow water bottom moored mine detection using the Navy s Comprehensive Acoustic Simulation System/Gaussian Ray Bundle (CASS/GRAB) model for a generic Very High Frequency (VHF) forward looking sonar. The effects of imprecise bottom type and wind speed data are evaluated to determine the impact of this variability on mine detection. The results show that signal excess variability is small and operational benefits may be maximized with a slightly better sonar. The effects of shallow and deep transducer placement in the water column are compared to determine which yields a greater probability of mine detection. The greatest probability of mine detection exists for the deep transducer. v

8 THIS PAGE INTENTIONALLY LEFT BLANK vi

9 TABLE OF CONTENTS I. INTRODUCTION... 1 II. CASS/GRAB MODEL... 7 A. RAY TRACING... 7 B. GAUSSIAN RAY BUNDLES C. CASS/GRAB CHARACTERISTICS III. ENVIRONMENTAL VARIABLIITY A. BOTTOM TYPE B. WIND SPEED IV. OPTIMIZING SIGNAL EXCESS A. TRANSDUCER DEPTH 5.18 METERS B. TRANSDUCER DEPTH 25 METERS C. SIGNIFICANCE OF DEPTH V. CONCLUSIONS APPENDIX A METER PEAK SIGNAL EXCESS APPENDIX B. 25 METER PEAK SIGNAL EXCESS APPENDIX C. CASS/GRAB MODEL INPUT CARD LIST OF REFERENCES BIBLIOGRAPHY INITIAL DISTRIBUTION LIST vii

10 THIS PAGE INTENTIONALLY LEFT BLANK viii

11 LIST OF FIGURES Figure 1. Mine placement in the water column (After N77, 2002) Figure 2. HF active sonar array mounted in the sail of the USS Asheville, a 688i Los Angeles Class submarine (From N77, 2002) Figure 3. UUV sonar systems and capabilities (From N77, 2002)... 4 Figure 4. Fan of acoustic rays (From NAVO, 1999)... 9 Figure 5. Simple Geometric Acoustics (From NAVO, 1999) Figure 6. CASS/GRAB Overview (From NAVO, 1999) Figure 7. Rays grouped into ray families (From NAVO, 1999) Figure 8. GRAB v1.0 flowchart (From NAVO, 1999) Figure 9. Muddy sand variability effects on signal Figure 10. Muddy sand variability effects on reverberation Figure 11. Muddy sand variability effects on signal excess Figure 12. Muddy sand variability histogram Figure 13. Sandy silt variability effects on signal Figure 14. Sandy silt variability effects on reverberation Figure 15. Sandy silt variability effects on signal excess Figure 16. Sandy silt variability histogram Figure 17. Muddy sand with wind variability effects on signal Figure 18. Muddy sand with wind variability effects on reverberation Figure 19. Muddy sand with wind variability effects on signal excess Figure 20. Muddy sand with wind variability histogram Figure 21. Sandy silt with wind variability effects on signal Figure 22. Sandy silt with wind variability effects on reverberation Figure 23. Sandy silt with wind variability effects on signal excess Figure 24. Sandy silt with wind variability histogram Figure 25. Signal excess curves at 5.18 m transducer depth and 300 m Figure 26. Signal excess curves at 5.18 m transducer depth and 600 m Figure 27. Signal excess curves at 5.18 m transducer depth and 900 m Figure 28. Signal excess curves at 5.18 m transducer depth and 1200 m Figure 29. Reverberation, signal and signal excess histograms at 5.18 m transducer depth Figure 30. Signal excess curves at 25 m transducer depth and 300 m Figure 31. Signal excess curves at 25 m transducer depth and 600 m Figure 32. Signal excess curves at 25 m transducer depth and 900 m Figure 33. Signal excess curves at 25 m transducer depth and 1200 m Figure 34. Reverberation, signal and signal excess histograms at 25 m transducer depth Figure m 25 m transducer signal excess difference at 300 m Figure m 25 m transducer signal excess difference at 600 m Figure m 25 m transducer signal excess difference at 900 m Figure m 25 m transducer signal excess difference at 1200 m ix

12 THIS PAGE INTENTIONALLY LEFT BLANK x

13 LIST OF TABLES Table 1. Bottom type geo-acoustic properties (From NAVO, 1999) xi

14 THIS PAGE INTENTIONALLY LEFT BLANK xii

15 ACKNOWLEDGMENTS The author is grateful for the professional expertise and guidance provided by Dr. Peter C. Chu and Mr. Steve D. Haeger. This study would not have been possible without the technical expertise of the following individuals: Chenwu Fan at the Naval Postgraduate School and Dr. Ruth E. Keenan at Science Applications International Corporation. This work was funded by the Naval Oceanographic Office. xiii

16 THIS PAGE INTENTIONALLY LEFT BLANK xiv

17 I. INTRODUCTION The United States military has undergone numerous changes since the end of the Cold War. Specifically, the US Navy experienced a shift in the area of engagement from the blue water (water depth greater than 100 m) Soviet threat, to littoral regions of the world. Sensors, tactics and platforms optimized to perform in a deep ocean, acoustically range independent environment operate inadequately in shallow, acoustically range dependent littoral regions. Littoral regions are reverberation limited and more complicated than the deep ocean. Most countries lack the robust Gross National Product of the US and have no intention of ever building a navy to oppose US Naval forces on the open seas in Mahanian fashion. The decision to focus naval military budgets on economical and lethal alternatives is prudent, as countries retreat to coastal defense postures. The weapons of choice are diesel submarines and sea mines, both of which present a credible threat to invading forces and require a disproportionately larger neutralizing force. Diesel submarines are very quiet, difficult to detect and a thrifty alternative to nuclear submarines. Technological advancements in battery design have resulted in higher capacity batteries with shorter recharge times. When employed for coastal defense, the long endurance advantage of nuclear submarines is negligible to these countries. Mines come in a multitude of variations and are readily available on the international market. Mines are designed to operate throughout the water column; on the surface, at various depths and on the bottom (Figure 1). There are an assortment of 1

18 actuators including contact, magnetic influence and acoustic. The incorporation of counters, where the mine detonates after a set number of trigger signals, can mask the presence of a minefield. Figure 1. Mine placement in the water column (After N77, 2002). In terms of cost effectiveness, mines are cheap to procure, deploy and require no upkeep once deployed. Any nation can acquire mass quantities of mines at far lower prices than shipbuilding. A single World War I Iranian contact mine in the Persian Gulf caused $96 million worth of damage to the USS Samuel B. Roberts (FFG-58) in April The return on investment is enormous, a cheap mine does significant damage and possibly removes an enemy ship from the theater. Mine deployment is uncomplicated and can be performed from basic surface craft. As evidenced by the Samuel B. Roberts, mines have the potential for long life spans without maintenance. The low target strength of mines combined with the complex littoral environment hinders minesweeping efforts. 2

19 Active sonar and unique High Frequency (HF) sonar systems along with Unmanned Undersea Vehicles (UUVs) are the new means to counter diesel submarines and mines. Active sonar has long been viewed by the submarine force with apprehension, since it gives away own ships position at twice the range of detection. New tactics for employing active sonar against diesel submarines have served to relieve deep rooted reservations and resulted in greater detection ranges. HF sonar generates higher resolution images necessary to distinguish mine-like objects from actual mines (Figure 2). Figure 2. HF active sonar array mounted in the sail of the USS Asheville, a 688i Los Angeles Class submarine (From N77, 2002). UUVs are essential to extend the sensor range of naval platforms and evaluate minefields without jeopardizing military lives. Specialized sonar systems, improved maneuverability and faster minefield assessment with the use of multiple UUVs make this a valuable asset to the Navy (Figure 3). The Navy s standard model for range dependent acoustic propagation, the Comprehensive Acoustic Simulation System (CASS) incorporates the Gaussian Ray Bundle (GRAB) eigenray model, in the 600 Hz to 100 khz frequency band (Keenan, 2000). This reverberation model works well to predict acoustic performance in the 3

20 littorals for signal excess given accurate inputs, such as bottom type, sound speed profile and wind speed. In 1980 the Generic Sonar Model evolved into CASS, consisting of system, acoustic and sonar analysis models (Weinberg, 2000). Figure 3. UUV sonar systems and capabilities (From N77, 2002). GRAB test ray amplitude is Gaussian distributed in depth, where energy is distributed over all depths. Test rays are sorted into families of comparable numbers of turning points and boundary interactions. Ray properties are then power averaged for each ray family to produce a representative eigenray of that family. CASS models monostatic and bistatic range dependent reverberation for transmitter/receiver to target cases. CASS evaluates reverberation as (Keenan, 2000): 4

21 d(reverb) = source level + 10log(scattering area) + scattering strength per unit area - transmission loss to scatterer + beam pattern of transmitter - transmission loss from scatterer + beam pattern of receiver. CASS computes signal excess (SE) by summing all eigenray path combinations for a range bin, then uses the peak signal to reverb/noise level. Fleet exercises are costly and access to foreign coastlines may not be politically feasible. In order to assess environmental and bottom type variability effects on SE in the littoral, the CASS/GRAB model is employed. Environmental predictions from the Naval Oceanographic Office (NAVO) and Fleet Numerical Meteorology and Oceanography Center (FNMOC) have inherent model errors associated with them. Many parts of the world have inadequate bottom type mapping in littoral regions and covert reconnaissance may be the only source of approximate bottom information. Bottom types can vary greatly in a small area and effect actual acoustic performance as bottom interaction changes. This thesis evaluates a generic VHF forward looking mine hunting sonar s performance in detecting a typical bottom mine. The analysis contained herein provides valuable insight as to how variability effects UUV and minesweeper performance. Extrapolations for diesel submarine detections can also be made from these findings. 5

22 THIS PAGE INTENTIONALLY LEFT BLANK 6

23 II. CASS/GRAB MODEL A. RAY TRACING CASS uses GRAB (Weinberg, 1983), as a range dependent propagation model to perform monostatic and bistatic active SE calculations. Ray tracing equations in rectangular coordinates from Snell s Law are given by: d ds dx p p =, ds x (1a) d ds dy p p =, ds y (1b) d ds dz p p =, (1c) ds z ds = ( dx) + ( dy) ( dz), (2) where (x, y) and z are the horizontal and vertical coordinates, and p is the acoustic pressure. Let r be the horizontal range and p 0 be the acoustic pressure at a reference distance ( r 0 ) from the source. The (r, z) plane is separated into triangular sections where the square of the gradient of slowness p is a constant, i.e., 7

24 p 2 p = r 2 2 p, constant. (3) z r = r d r c. (4) z = z d z c. (5) Range and depth variation, r and z respectively, of a ray is obtained by r dr p ds c + dr p ds d 1 = z dz p ds c + dz p ds d 1 = h, (6) dr dr r = h p + p, (7) ds c ds d dz dz z = h p + p. (8) ds c ds d where c and d are the initial and final points, respectively. Let the v-th ray be launched at the source with the angle ( θ v, 0 ), cross range r at depth z v, and travel time T v (Figure 4). The horizontal and vertical slowness is represented by p cosθ v r, v =, (9) cv and p sinθ v z, v =. (10) cv where c v is the sound speed, and θ v the horizontal inclination angle. The travel time ( T z, v ) at the field-point depth z is computed from T v by 8

25 T ( z z ) z, v = Tv + pz, v v. (11) Figure 4. Fan of acoustic rays (From NAVO, 1999). Volume attenuation and boundary reflection losses are simulated by the pressure ratio Γ v and phase shift Φ v, respectively. Rayleigh bottom forward loss is determined from the first three columns of the geo-acoustic table for various bottom types (Table 1). Duplicate grain size indexes are listed to cover all commonly used geo-acoustic bottom types. Sand is the default bottom type for CASS/GRAB. CASS utilizes all six geoacoustic parameters listed to calculate backscattering strength. Γ = Πγ Πγ, (12) v s, v b, v Φ v = φ + φ. (13) x, v b, v 9

26 where γ s,v is the surface reflection coefficient amplitude; γ b,v is the bottom reflection coefficient amplitude; φ s,v and φb,v are the surface and bottom phase shifts Different from the classical ray theory, the Gaussian ray bundle amplitudes are global and distributed to some degree for all depths, where classic ray path amplitudes are local. For other quantities of source angle, target angle, travel time and phase, Gaussian ray bundles match classic rays. Index Table 1. g/cm Bottom type geo-acoustic properties (From NAVO, 1999). 10

27 B. GAUSSIAN RAY BUNDLES The source dependent factor β v, 0, is a conversion term to equate energy within a Gaussian ray bundle with a geometric-acoustic ray tube. Gaussian ray bundle amplitude of the power at target depth z of the v -th test ray is Ψ v. The conservation of energy for simple geometric acoustics states that energy in a ray tube is equivalent to the energy in the ray tube at a reference range (Figure 5). β 2 2 v, 0 0 r, v,0 v,0p0 = r p θ, (14) Ψ v = z z exp 0.5 r σ 2 βv,0γv v 2πσ v pr, v v 2, (15) I A = I. (16) o A 0 where σ v is the standard deviation. The Gaussian eigenray the ray family. Gaussian eigenray amplitude Ψ e, is derived by the power addition of the ray bundles in A e, is the square root of the eigenray. Ψe = Ψ v, (17) v Ae = Ψ. (18) e 11

28 Figure 5. Simple Geometric Acoustics (From NAVO, 1999). Additional eigenray properties are obtained from the power weighted averages of the ray bundle properties, such as the source angle, θ = Ψ e 1 e v Ψvθ v,0, (19) 12

29 the horizontal slowness, 1 p r, e = Ψe Ψv pr, v v, (20) the vertical slowness, 1 p z, e = Ψe Ψv pz, v v, (21) the boundary phase shift, Φ e = Ψ 1 Ψ Φ, (22) e v v v and travel time, 1 Ψ T e = Ψe vtz, v v. (23) GRAB contains sound speed conversion models such as Leroy s equation (Leroy, 1969) and Millero-Li s equation (Millero, 1994), which is an adjustment to the original Chen-Millero equation (Chen, 1977). The Navy Standard model is Wilson s second equation for temperature-salinity-sound speed conversion (Wilson, 1960). GRAB defaults to Leroy s equation for sound speed conversions, where numerically stable polynomials are fit to Wilson s data. Propagation losses are calculated as coherent and random pressure fields based on Gaussian eigenray amplitude and discrete phase shifts P e, as a complex value, Φ e, which yield eigenray pressure P = A exp( iω Τ + iφ ), (24) e e e e 13

30 P coherent = P e, (25) e 2 P random = P e. (26) e C. CASS/GRAB CHARACTERISTICS NUWC Code 842 validated and adapted GRAB for their weapons program in CASS can use any of four incorporated propagation models in its calculations, including FAME and COLOSSUS (Figure 6). CASS/GRAB has been certified for use in the frequency band 600 Hz to 100 khz. Figure 6. CASS/GRAB Overview (From NAVO, 1999) 14

31 GRAB traces a user-specified fan of test rays through a range dependent environment while identifying ray properties of travel time, amplitude, phase, source angle, ray type and target angle. GRAB groups these rays into ray families of adjacent rays with comparable travel paths based on the number of surface and bottom interactions (Figure 7) and power averages the rays in the ray family to form up to two eigenrays representative of each family. From these representative eigenrays, coherent or random propagation loss is calculated (Figure 8). Figure 7. Rays grouped into ray families (From NAVO, 1999). 15

32 Figure 8. GRAB v1.0 flowchart (From NAVO, 1999). 16

33 III. ENVIRONMENTAL VARIABLIITY A. BOTTOM TYPE The lack of detailed bottom type data in numerous hot spots of the world due to political restrictions and shrinking monetary resources reduces model accuracy and degrades accurate planning timelines. The effects of small variations in bottom type inputs are examined to assess the impact on model predictions and detection ranges. It is important to evaluate input sensitivity for credible time and range estimates of minesweeping or diesel submarine searches based on limited data. The first bottom type variation examined is muddy sand, which corresponds to grain size index 3.0 (from Table 1). The variation is from grain size index , in 0.5 grain size index increments, corresponding to bottom types ranging from muddy gravel to clayey sand. The parameters are wind speed of 10 kts (5.14 m/s), tilt angle four degrees down, bottom depth 30 m, transducer depth 5.18 m (17 feet) and a range of evaluation out to 1200 m for a mine-like object on the sea bottom. It should be noted CASS/GRAB identifies tilt angles up as negative and tilt angles down as positive. However, tilt angles hereafter will be labeled positive for upward and negative for downward to alleviate any confusion with common nomenclature. The signal is erratic below 100 m with 15 db signal changes, peaking at 90 db around 130 m, and then decaying to 38 db at 1200 m (Figure 9). The signal variability at the peak is 2 db, gradually converging to less than or equal to 1 db as the range increases. 17

34 Figure 9. Muddy sand variability effects on signal. Reverberation contains greater variability than the signal; it also diverges as range increases (Figure 10). The reverberation is less irregular closer than 100 m, peaking at 90 db about 130 m, then shrinking to roughly 47 db at 1200 m. The reverberation variability is 3 db at the peak, diminishing to 1 db at 250 m, then diverging to 4 db at 1200 m. Noise is modeled as a constant value for all of the cases over the entire range and has no variability since the wind speed is constant. As such, no graph is given. 18

35 Figure 10. Muddy sand variability effects on reverberation. The signal excess is erratic for ranges less than 100 m with 15 db changes, building to a peak at roughly 200 m of 4 db, then decreases with range out to 1200 m to 14 db (Figure 11). The variability of signal excess at the peak is about 3.5 db, decreasing slightly to 3 db until 800 m, and then converging to 1 db variability at 1200 m. 19

36 Figure 11. Muddy sand variability effects on signal excess. The histogram of muddy sand illustrates the peak signal excess of 4 db, and the majority of signal excess lies between 5 and 7 db (Figure 12). This is due to the plateau in the signal excess graph between 600 and 900 m. The left skewed shape is due to the early signal excess peak at 200 m then the gradual decay as range increases and values less than 15 db are due to erratic signal excess at close range. 20

37 Figure 12. Muddy sand variability histogram. The effects of muddy sand variability are important if 3 db will significantly effect the particular sonar performance for the desired detection ranges. Employment of better sonar equipment with a five to 7 db gain will significantly enhance detection ranges for this scenario from an optimal 350 m out to 900 m. The last bottom type variation examined is sandy silt, corresponding to grain size index 5.0 (from Table 1). Sandy silt is varied from clayey sand to silt, over the grain size index , in 0.5 grain size index increments. All the parameters of wind speed, tilt angle, bottom depth, transducer depth and range are the same as the first run. 21

38 The signal is still erratic below 100 m but with smaller 10 db signal changes, a lower peak of 88 db at nearly 130 m, decaying to 38 db at 1200 m (Figure 13). There is greater signal variability with 3 db at the peak, growing to 4 db until about 700 m, and then converging to 2 db. The reverberation still contains greater variability than the signal (Figure 14). Reverberation is more consistent than signal less than 100 m, the peak of 91 db is at 130 m, then lessening to about 43 db at 1200 m. Reverberation variability is 3 db at the peak, growing to 6 db at 1200 m. Figure 13. Sandy silt variability effects on signal. 22

39 Figure 14. Sandy silt variability effects on reverberation. The signal excess less than 100 m has a max fluctuation of 18 db, reaching a peak of 2 db at 250 m, and then shrinking to 14 db at 1200 m (Figure 15). Signal excess variability at the peak is 6 db, contracting to 5 db until 750 m, and then reducing to 1.5 db at 1200 m. 23

40 Figure 15. Sandy silt variability effects on signal excess. The sandy silt variability histogram demonstrates the signal excess peak of 2 db, and the bulk of signal excess subsides between 1 and 4 db (Figure 16). The plateau between 400 and 800 m in the signal excess graph explains this. The left skewed shape is due to the early peak in signal excess followed by the gradual signal excess reduction with range and values less than 15 db are due to erratic signal excess at close range. 24

41 Figure 16. Sandy silt variability histogram. The variability of sandy silt effects is important if 6 db will have a substantial impact on sonar performance. The histogram illustrates that a one to 4 db gain significantly boosts detection ranges from a mere 300 m out to 850 m for a best case scenario. Comparing muddy sand and sandy silt variability histograms, it can be noted that the muddy sand variation has a greater overall change in signal excess over its range. However, the histogram of sandy silt portrays the lower standard deviation and the signal excess peak count lies at a higher value. The sandy silt signal excess maximum is lower, 25

42 but does not vary with range as fast as muddy sand since it has a lower slope to its curve. For ranges closer than 400 m, sandy silt contains more sensitivity to input variations; while muddy sand exhibits greater input sensitivity for ranges greater than 600 m. B. WIND SPEED Small errors in wind speed due to meteorological model inaccuracies or inadequate accuracy in wind observations can effect acoustic model accuracy and operational timing. The effects of a small wind speed variation are examined based on a nominal 10 kt wind speed. Two different bottom types are examined, muddy sand and sandy silt with wind speed kts ( m/s). The CASS/GRAB input parameters are tilt 4, bottom depth 30 m, transducer depth 5.18 m and range out to 1200 m based on a minelike object on the bottom. The muddy sand with wind variation signal graph has virtually no inconsistency until about 450 m, where it diverges to a 5 db variation at 1200 m (Figure 17). The signal fluctuates below 100 m, and then rises to the signal peak of 89 db at 130 m, followed by the fade in signal strength down to 40 db at 1200 m. Reverberation contains more variability over its entire range, with an early peak at 50 m containing a 16 db variation, converging to 4 db as wind speeds under 10 kts build to another maximum (Figure 18). After 230 m the reverberation diverges again to a peak of 9 db at 330 m, and then slowly converges to 4 db at 1200 m. Reverberation demonstrates less fluctuation at close ranges with a peak of 98 db at 50 m, then decreasing with range to 46 db at 1200 m. It can be derived from the muddy sand wind 26

43 variability reverberation graph, that lower wind speeds shift the reverberation maximum to farther ranges and lower magnitudes. An interesting feature is at 20 m, the wind speed variations converge. Wind speeds below 7 kts reach a local maximum, while wind speeds greater than 7 kts reach a local minimum. Figure 17. Muddy sand with wind variability effects on signal. Signal excess oscillates below 100 m, peaking at 6 db around 200 m, and then decays to 13 db at 1200 m (Figure 19). The signal excess variability is relatively large, as it changes from 4.5 db at the peak, diverging to about 9 db from 320 m to 800 m, and then converging to 6 db at 1200 m. 27

44 Figure 18. Muddy sand with wind variability effects on reverberation. Noise is modeled as a constant value for each wind speed, increasing with greater wind speed. The contribution is small, about 0.1 db for each 1 kt increase and a total 1 db over the wind speed variation. This plays a small role in generating greater signal excess variation. The histogram of muddy sand wind variability reveals the majority of signal excess subsides between 2 and 11 db (Figure 20). The reason for this trait is the plateau and 9 db variation in the signal excess plot between 500 and 800 m. The signal 28

45 excess after 90 m never falls below 16 db, therefore, the left skewed values are due to erratic behavior at close range. Figure 19. Muddy sand with wind variability effects on signal excess. The effects of wind speed variation for muddy sand is significant since 9 db departures occur over one-third of the range and 6 db or more for ranges greater than 250 ms. Better sonar equipment with a gain of three or more decibels will significantly increase detection ranges from a crest of 530 ms to about 900 ms for a best case scenario. 29

46 Figure 20. Muddy sand with wind variability histogram. The last wind speed variation is for a sandy silt type bottom with the same model input parameters as the previous case. There is still virtually no inconsistency initially but extending farther out to about 700 m, the signal then abruptly jumps to a 3 db variation that exists until 1200 m. The fluctuation in signal strength still exists below 100 m, building to a lower peak of 84 db at 130 m, followed by the decay in signal strength with range to 38 db at 1200 m (Figure 21). 30

47 Figure 21. Sandy silt with wind variability effects on signal. The reverberation still contains more variability over its entire range, with the peak at 50 m and 16 db variation, which rapidly converges to no variation at 170 m as wind speeds below 10 kts reach another maximum and then diverging again at 300 m out to 8 db at 1200 m (Figure 22). At close ranges reverberation demonstrates small oscillations building to a lower peak of 97 db at 50 m and then dropping off to a lower value of 39 db at 1200 m. There is still an interesting sharp convergence as wind speeds below 7 kts reach a local maximum at 20 m and wind speeds above 7 kts reach a local minimum. It should be noted that lower wind speeds shift the reverberation maximum to a higher range and lower magnitude. Upon closer inspection of the randomly distributed 31

48 wind variation runs, the sandy silt wind variations did not generate any wind speeds of 6 or 15 kts. This has a negligible effect on the sandy silt plots, while the reverberation peak would be about 2 db higher from interpolation. The signal excess still oscillates below 100 m; however, the peak of 2 db is reached much later at 750 m, and then decays to 14 db at 1200 m (Figure 23). The signal excess variability is smaller overall, nearly vanishing at 230 m, then diverging to 6 db at the peak, followed by a slow convergence to 5 db at 1200 m. Figure 22. Sandy silt with wind variability effects on reverberation. 32

49 Figure 23. Sandy silt with wind variability effects on signal excess. The noise contribution to variation is still minor, with about 1 db over the entire wind speed variation and 0.1 db for each additional knot of wind speed. The sandy silt histogram exhibits a peaky nature with a smaller standard deviation where the majority of signal excess lies between 1 and 6 db (Figure 24). This characteristic is due to the plateau from 200 to 800 m. The left skewed results less than 16 db are due to close range oscillations below 100 m. 33

50 Figure 24. Sandy silt with wind variability histogram. The effects of wind speed variation for sandy silt are less than that of muddy sand, with a peak variation of 6 db over a short distance and 5 db variations for less than half the range. Sonar equipment with an improved gain of 1 to 6 db would significantly enhance detection range by widening the window of detection. The variations for sandy silt are most significant closer than 100 m and greater than 750 m, for ranges that lie outside of this, variations are less than 5 db. Comparing muddy sand and sandy silt wind speed variations, the muddy sand demonstrates greater input sensitivity to wind speed variations over its entire range. 34

51 Wind speed accuracy is therefore more important to accurate acoustic modeling for muddy sand. For sandy silt, wind speed variation becomes 2 db more important for ranges greater than 750 m. 35

52 THIS PAGE INTENTIONALLY LEFT BLANK 36

53 IV. OPTIMIZING SIGNAL EXCESS A. TRANSDUCER DEPTH 5.18 METERS Utilizing the following graphs, the optimal tilt angle for the present environment can be selected to maximize signal excess at a given range. The signal excess is examined in 300 m increments out to 1200 m for tilt angles + 4 to 12 in four degree increments. Wind speeds from 5 to 25 kts (2.57 to m/s) in 5 kt steps for bottom type grain size indexes 0.5 (coarse sand) to 6.0 (silt) in 0.5 increments, in a 30 m water depth are analyzed. Investigating the signal excess at 300 m there are some interesting features to note (Figure 25). There is a dip in signal excess at bottom type grain size index 5.0 (sandy silt) for all wind speeds and tilt angles down to 8. For coarse sand to clayey sand, tilt angle 4 presents the highest signal excess for all wind speeds. For tilt angles 4 and down, the finer bottom types, sandy silt to silt, exhibit little sensitivity to wind speed with only a 3 db difference. Also, for bottom type grain size indexes 4.0 (clayey sand) to silt, tilt angles 4 and down offers higher signal excess for all wind speeds. The signal excess also experiences an elevated value for the highest wind speeds at these fine bottom types for tilt angles 8 and 12. Wind speed has the greatest effect on muddy gravel to clayey sand for all tilt angles with a peak difference of 14 db. The highest wind speeds, 20 and 25 kts (10.29 and m/s), display comparable signal excess for all tilt angles and bottom types. The character of the signal excess curves change at 600 m (Figure 26). The finer bottom types, sandy silt to silt still exhibit little sensitivity to wind speed for tilt angles 37

54 4 and down. However, the peak signal excess values are all shifted to the finer bottom types. The greatest sensitivity to wind speed variation exists for medium bottom types, muddy gravel to clayey sand, at tilt angles of 0 and to a lesser degree for all other tilt angles. The coarse bottom types, coarse sand to muddy gravel, demonstrate maximum signal excess at tilt angle 0. Wind speeds of 20 and 25 kts continue to display comparable signal excess for all tilt angles and bottom types. Figure 25. Signal excess curves at 5.18 m transducer depth and 300 m. 38

55 Figure 26. Signal excess curves at 5.18 m transducer depth and 600 m. The character of the signal excess curves change again at 900 m (Figure 27). For tilt angles down to 4, the signal excess builds to a maximum for finer bottom types. A tilt angle of 0 offers the best signal excess for all bottom types and wind speeds, slightly better than 4 tilt angles. For tilt angles 8 and 12, the signal excess diminishes for the fine bottom types, after building to a peak for medium bottom types. The greatest signal excess variations occur for tilt angle 12 for the medium bottom types, with a peak value of 14 db difference for very fine sand. Tilt angle 12 offers the lowest signal excess for all wind speeds and bottom types. Wind speeds of 20 and 25 kts show little variation for bottom type or tilt angle and maintain similar values of signal excess. 39

56 Figure 27. Signal excess curves at 5.18 m transducer depth and 900 m. The character of the signal excess curves does not change much from 900 m, at 1200 m (Figure 28). A tilt angle of 0 still offers the best signal excess for all bottom types and wind speeds. Tilt angle 12 contains the greatest signal excess variation of 20 db for medium bottom types and the lowest signal excess for all bottom types and wind speeds. For tilt angles down to 4, each wind speed shows little variation in signal excess magnitude from muddy gravel to silt. 40

57 Figure 28. Signal excess curves at 5.18 m transducer depth and 1200 m. The histograms of reverberation, signal, and signal excess are for all ranges out to 1200 m, tilt angles + 4 to 12, bottom types gravelly sand to silt, wind speeds 5 to 25 kts, water depth 30 m and transducer depth 5.18 m (Figure 29). The peak of signal excess is at 7 db, with a few instances of readings as high as + 7 db. The signal excess histogram demonstrates a nearly Gaussian distribution with a left skewed shape due to the farthest ranges and tilt angles of 12 effects. The signal histogram exhibits a bimodal distribution, centered at about 48 and 75 db with slight left and right skewness. The reverberation histogram also appears to have a slight bimodal distribution, centered at about 50 and 78 db, nearly matching the signal histogram modes. 41

58 Figure 29. Reverberation, signal and signal excess histograms at 5.18 m transducer depth. B. TRANSDUCER DEPTH 25 METERS The following graphs depict the same conditions of range, tilt angles, bottom types and water depth as the previous section, but for a deeper transducer at 25 m. Examining the signal excess at 300 m reveals the largest wind sensitivity at + 4 tilt angle, with up to 12 db difference for medium and fine bottom types (Figure 30). Coarse and medium bottom types demonstrate maximum signal excess at tilt angle 0 for all wind speeds. Fine bottom types exhibit the greatest signal excess at 4 tilt angle for all wind speeds. Tilt angle 12 shows a significant signal excess loss for fine bottom types at all 42

59 wind speeds. The highest wind speeds display no difference in signal excess for all bottom types down to 8 tilt angle. Figure 30. Signal excess curves at 25 m transducer depth and 300 m. The signal excess at 600 m contains the largest wind variability at tilt angle + 4 again, with a variance up to 13 db (Figure 31). The coarse and medium bottom types continue to be optimized at tilt angle 0 for all wind speeds. Tilt angle 4 still encloses the utmost signal excess for fine bottom types of all wind speeds. Tilt angle 12 has the worst signal excess for all wind speeds with a momentous reduction for fine bottom types. This is due to greater bottom interaction with a deeper transducer depth and 5 m to 43

60 the sea bottom. The highest wind speeds still match down to 8 tilt angle; but at 12, 25 kt winds show a marked reduction in signal excess for all bottom types except for the coarsest bottoms. Figure 31. Signal excess curves at 25 m transducer depth and 600 m. The signal excess for 12 tilt angle at 900 m displays marked degradation for all bottom types, but especially for fine bottom types (Figure 32). The optimal tilt angle for all bottom types is 0 to maximize signal excess. Tilt angles down to 4 generate similar signal excess curves, with slight magnitude variations. Tilt angles down to 8 favor fine 44

61 bottom types to generate the most signal excess. The highest wind speeds remain nearly identical for all bottom types. Figure 32. Signal excess curves at 25 m transducer depth and 900 m. Tilt angle 0 still provides the highest signal excess for all bottom types and wind speeds at 1200 m (Figure 33). Tilt angles down to 4, in spite of everything, offer similar signal excess with only a small difference in magnitude. Tilt angle + 4 offers the greatest sensitivity to wind speed for medium bottom types. The highest wind speeds match shapes, but diverge slightly more than before. Tilt angle 12 is of no use at this range, it contains the worst signal excess of all the graphs. 45

62 Figure 33. Signal excess curves at 25 m transducer depth and 1200 m. The histograms of reverberation, signal, and signal excess are for all ranges out to 1200 m, tilt angles + 4 to 12, bottom types gravelly sand to silt, wind speeds five to 25 kts, water depth 30 m and transducer depth 25 m (Figure 34). The signal excess is bimodal, with peaks at 7 db and + 12 db; the strong left skewness is due to tilt angles at 12. The signal histogram contains a plateau at the peak around 50 db, which may contain a small bimodal peak about 75 db, with a gradual right skew. The reverberation histogram appears to contain a minor bimodal distribution with peaks at 50 and 72 db. The standard deviation is smaller, contributing to a narrower distribution of a higher 46

63 magnitude than the signal. This yields a signal excess histogram dominated by negative values. Figure 34. Reverberation, signal and signal excess histograms at 25 m transducer depth. C. SIGNIFICANCE OF DEPTH Optimizing signal excess in range increments of 300 m for the two transducer depths of 5.18 m and 25 m, in 30 m of water yields results similar to the histogram findings. At 300 m the shallow transducer has a slight advantage up to 3 db for the higher wind speeds of 15, 20 and 25 kts for coarse and medium bottom types (Figure 35). 47

64 For all other cases, the deep transducer is superior with up to a 12.5 db advantage. The 25 m transducer possesses the greatest probability of detection at this range, with the highest positive signal excess values. Figure m 25 m transducer signal excess difference at 300 m. At 600 m the coarse and medium bottom types have similar signal excess with less than a 1 db difference for virtually all wind speeds at the two transducer depths (Figure 36). The exception is for 15 kts, where the shallow transducer has a slight advantage of 1.5 db and up to nearly 3 db for fine bottom types. The two highest wind speeds still favor the 5.18 m transducer with a small signal excess improvement. At 5 kts 48

65 winds speed, the 25 m transducer has a 1.5 db advantage for coarse sand and over 3 db for sandy silt to silt. The deep transducer holds a slight advantage for 10 kts wind speed, building to 2.5 db for silt. The 25 m transducer still possesses the greatest probability of detection at this range, with the highest positive signal excess values except for clayey sand and coarse silt; where the shallow transducer has a meager quarter decibel and half decibel advantage, respectively. Figure m 25 m transducer signal excess difference at 600 m. The 5.18 m transducer contains better signal excess for nearly all cases at 900 m (Figure 37). The deep transducer dominates for the two highest wind speeds with coarse 49

66 silt to silt bottoms. It should be noted there are no positive signal excess values at this range, a better sensor would be required to make detections. Figure m 25 m transducer signal excess difference at 900 m. The shallow transducer encompasses the greatest signal excess for all cases, with over 4 db advantage for 10 kts wind speed (Figure 38). The greatest signal excess at this range is less than 9 db for 5 kts wind speed. Therefore, the present sensor would not detect a mine-like bottom moored object at this range. 50

67 Figure m 25 m transducer signal excess difference at 1200 m. 51

68 THIS PAGE INTENTIONALLY LEFT BLANK 52

69 V. CONCLUSIONS The muddy sand variability was approximately 3 db for ranges less than 800 m and smaller than the sandy silt variation of 5 db; however, for ranges greater than 800 m, the sandy silt was less variable than muddy sandy. For ranges at the peak signal excess, sandy silt contained the greatest variability of 6 db. For the present sensor analyzed at 5.18 m, this makes the difference between detection and no detection in sandy silt areas. The importance of this finding is that for this type of bottom mine and sensor, the model may predict an identifiable environment; while the actual environment would yield no positive signal excess and hence no detections. This worst case scenario indicates that bottom type data is more significant for fine bottom types. The deadly results of this would be an area certified clear of mines, that still contains mines. For the present conditions, the muddy sand variability has a small effect on detection since there is a positive signal excess for all cases. Therefore, for small variations of muddy sand, the impact on detection is minute. Employment of a better sensor, if available, would offset the sandy silt variability effect due to inadequate data. Evaluation of these signal excess graphs reveals that a better sensor with 1 to 4 db gain significantly boosts detection ranges and probability, in spite of the greater variability present. This is due to the fact that sandy silt has a lower signal excess slope than muddy sand out to 800 m. The effects of wind variability on muddy sand and sandy silt are just the opposite of bottom type variability. Muddy sand now has the greatest variability up to 9 db, while sandy silt only reaches 6 db. Muddy sand is more sensitive to wind variability than 53

70 sandy silt over the entire range. Despite the greater variability disadvantage, muddy sand still demonstrates positive signal excess for all variations of wind speed. Sandy silt is less sensitive to wind variability, however, this fine bottom type exhibits low signal excess. Only the lowest wind speeds generate positive signal excess from 400 to 800 m, in this region the signal excess slope is nearly zero. For the present sensor and bottom mine, the wind speed accuracy is extremely important. If wind speed prediction and uncertainty exceed 7 kts, mine detection is unlikely for the current sensor. If the current sensor must be used, predicted wind and uncertainty must remain below this threshold for sandy silt. Should another sensor be available, only a 3 db gain would allow mine detection for all wind variations. The generic VHF forward looking mine hunting sonar performance for muddy sand is satisfactory; however, sandy silt requires greater bottom type and wind speed accuracy to determine if detections are likely. In the event higher accuracy bottom type and wind speed data are unattainable for sandy silt bottom types, alternative sensors or tactics are recommended to overcome this sensors shortcomings. The findings indicate that improved sensor performance of a few decibels far out weigh the benefits of higher accuracy bottom and wind data. Higher accuracy inputs and improved sensors are expensive, as such; money would be better spent on sensor improvements for a greater return on investment. Mine hunters and UUVs can vary the depth and tilt angle the transducer uses to search for mines. The analysis reveals that medium bottom types are the most sensitive to wind speed variations for the tilt angles and transducer depths examined. Variations up to 15 db are demonstrated for wind speed variations of kts. Medium bottom 54

71 types demonstrate greater signal excess at ranges under 300 m, while fine bottom types peak at farther ranges. At ranges greater than 600 m, tilt angles of 12 provide insufficient signal excess to be useful. The highest wind speeds of 20 and 25 kts have nearly the same signal excess for all tilt angles and bottom types at transducer depth 5.18 m. Transducers at 25 m exhibit the same characteristics with a few exceptions at tilt angle 12 at 600 m and closer. This indicates that wind speed accuracy for the highest wind speeds examined are inconsequential. The histograms of signal excess confirm that transducers at 25 m generate substantial signal excess up to 23 db compared to 7 db for 5.18 m transducers. The deep transducer histogram also contains a larger area of positive signal excess, indicating a greater probability of bottom moored mine-like detection. The plots of shallow versus deep transducer signal excess demonstrate the advantages of greater detection using a deep transducer for the majority of cases. Therefore, to increase the probability of bottom mine detection by a mine hunter or UUV utilizing this generic VHF forward looking sonar, placement of the transducer deeper in the water column is advantageous. 55

72 THIS PAGE INTENTIONALLY LEFT BLANK 56

73 APPENDIX A METER PEAK SIGNAL EXCESS 57

74 58

75 59

76 60

77 APPENDIX B. 25 METER PEAK SIGNAL EXCESS 61

78 62

79 63

80 64

81 APPENDIX C. CASS/GRAB MODEL INPUT CARD OUTPUT FILE = OUTPUT RESET OUTPUT DEVICE INPUT FILE = SVP ADD INPUT FILE BOTTOM DEPTH = 27 M WIND SPEED = 5 KNOTS BOTTOM SEDIMENT GRAIN SIZE INDEX = 0.5 SOURCE DEPTH = 17 FT TRANSMITTER DEPTH = 17 FT INPUT FILE ADD INPUT FILE = BIZONAL FREQUENCY MINIMUM = HZ FREQUENCY MAXIMUM = HZ RADIUS OF CURVATURE = KM AMBIENT NOISE SPECTRUM MODEL = WENZ SELF NOISE SPECTRUM MODEL = TABLE SELF NOISE SPECTRUM TABLE = 40 DB AMBIENT NOISE COMPONENT = TOTAL SOURCE LEVEL MODEL = TABLE SOURCE LEVEL TABLE = 200 DB PULSE LENGTH = 0.30 MS TIME MINIMUM =.02 S TIME MAXIMUM = 1.70 S TIME INCREMENT = 0.15 MS 65

82 RANGE MINIMUM = M RANGE MAXIMUM = 1.20 KM RANGE INCREMENT = M AMBIENT NOISE THRESHOLD MODEL = TABLE AMBIENT NOISE THRESHOLD TABLE = 10.5 DB REVERBERATION THRESHOLD MODEL = TABLE REVERBERATION THRESHOLD TABLE = 10.5 DB TARGET STRENGTH MODEL = FREQUENCY TARGET STRENGTH TABLE = -20 DB ADDITIONAL INFORMATION = SRN DEPTH UNIT = M RANGE UNIT = M TRANSMITTER TILT ANGLE = 0 DEG VERTICAL ANGLE MINIMUM = DEG VERTICAL ANGLE MAXIMUM = 60.0 DEG VERTICAL ANGLE INCREMENT = 0.1 DEG SURFACE SCATTERING STRENGTH MODEL = APL/UW SURFACE REFLECTION COEFFICIENT MODEL = APL/UW BOTTOM REFLECTION COEFFICIENT MODEL = RAYLEIGH BOTTOM SCATTERING STRENGTH MODEL = APL/UW OUTPUT FILE RESET OUTPUT DEVICE RECEIVER TILT ANGLE INPUT FILE ADD INPUT FILE = SVP11111 = SETILT = -4 DEG OUTPUT FILE RESET OUTPUT DEVICE RECEIVER TILT ANGLE = SVP11112 = 0 DEG 66