Resonance classification of swimbladder-bearing fish using broadband acoustics: 1-6 khz Tim Stanton The team: WHOI Dezhang Chu Josh Eaton Brian Guest Cindy Sellers Tim Stanton NOAA/NEFSC Mike Jech Francene Stroman NRL Joe Fialkowski Roger Gauss Ed Kunz Richard Menis NPS Ben Jones Independent Aija Irene Briga A.C. Hill
Interpreting acoustic data in terms of meaningful biological quantities Echogram Echo Voltage Scattering models Length, type, Numerical density, Spatial/temporal distribution
Resonance classification using broadband acoustics Swimbladder Resonance One Fish Broadband Scattering dominated by swimbladder, which resonates at low frequencies Narrowband Depth Resonance depends on: Fish size and depth Mixed assemblage
Types of acoustic signals: Narrowband gated sine wave Broadband Impulse Linear frequency modulation chirp!! time frequency
Pulse Compression Processing Method: Cross correlate broadband echo with replica (similar to matched filter processing) Result: Short signal, duration = (bandwidth) -1 Improved resolution and SNR through processing
Advantages of broadband signal over narrowband signal 1. Broad spectrum provides more information (for classification of marine organisms) Through signal processing, one can achieve: 2. Higher range resolution 3. Higher signal-to-noise ratio
Ocean experiment two mid-frequency systems ~3 kts Horizontal-looking ( 1.5-9.5 khz (NRL) HLA mode < ~ 10 km) Broadband signals time frequency Down-looking ( < ~ 200 m) 1.0 6 khz (WHOI)
Area of study
WHOI down-looking, modified-echosounder system 4-Channel System 1-6 khz 4-20 khz 30-70 khz 50-110 khz Calibration: at sea Standard target (sphere) Eliminated sphere resonances Stanton/Chu JASA (2008)
Pelagic trawl FR/V Delaware II Atlantic herring
Net samples all combined 2008 Atlantic herring Red fish Number of fish Silver hake Butterfish Pearlsides Red hake Length (cm) Squid 0 10 20 30 40
Traditional system (120 khz narrowband) New system (lower frequency, broadband) Improved resolution: Broad bandwidth, matched filter processing and flying low Reduced ambiguity: Lower frequency naturally selects only swimbladder-bearing fish Stanton/Chu/Jech/Irish ICES JMS (2010)
Traditional system (120 khz narrowband) New system (lower frequency, broadband) Improved resolution: Broad bandwidth, matched filter processing and flying low Reduced ambiguity: Lower frequency naturally selects only swimbladder-bearing fish Stanton/Chu/Jech/Irish ICES JMS (2010)
Resonance Classification Resonance (3.7 khz) Atlantic Herring Dense Patch (0.3 m -3 ) Sparse Patch (0.05 m -3 ) 2 100 Observations: Strong echo (upper) and weak echo (lower) have same resonance freq. Reduced ambiguities: Difference in echo strength is due to difference in density of fish, not size of fish or its orientation distribution Stanton/Chu/Jech/Irish ICES JMS (2010)
Mixed assemblages of fish Silver Hake Herring 25 cm Herring 25 cm Mix of fish 2-5 cm Stanton/Sellers/Jech CJFAS (2012)
Resonance classification of mixed assemblages Aggregation J Aggregation E Depth (m) Lower resonance Lower resonance Larger fish Higher resonance Higher resonance Smaller fish Distance (km) Distance (km) Stanton/Sellers/Jech CJFAS (2012)
NRL horizontal-looking long-range system RECEIVER SOURCE HLA and VLA configurations 32 elements cut for 5 khz Vertical Line Array (VLA) cut for 3 khz Frequency: 1.5-9.5 khz 10 transducers Towable at up to 4 kts 2 sensor suites (depth, tilt, etc.) ~Omni azimuthally Source level (SL) best: ~3.5-9 khz Waveforms and Down-range resolutions 2.5-3.5 khz LFM: 6.0-1.5 khz LFM: 1.5-9.5 khz LFM: 0.77 m 0.17 m 0.10 m
50 50 Predictions sourc receiver e 95 90 Data Fish ensonified in 1 st two annuli 85 80 150 150 75 70 1470 1510 0 1 2 3 4 5 6 7 8 9 10 m/s non-herring: 30 m above seafloor km herring: 140-190m 65 Fish ensonified in deep shadow zone? Or shallow refracted path? Fish ensonified in third annulus 18
Observed Echo Spectra Observed Spectra (db) 48 Set A redfish 46 44 42 40 38 N echoes = 194-32 -34-36 -38-40 -42 54 52 50 48 46 44 redfish herring Set B N echoes = 74-32 -34-36 -38-40 -42 Modeled VSS (db) Observed Spectra (db) 48 redfish herring 46 44 42 40 Set C -32 N echoes = 228-34 -36-38 -40 38-42 1 2 3 4 5 frequency (khz) 40 redfish herring 38 36 34 32 Set D -32 N echoes = 381-34 -36-38 -40-42 30 1 2 3 4 5 frequency (khz) Modeled VSS (db) Jones (2012) and Jones et al. (in manuscript) 19
Summary Range resolution: Improved through broadband processing Signal-to-noise ratio: Improved through broadband processing Ambiguities: Reduced through resonance/spectral classification, use of low frequencies Multiple size classes: Can spectrally resolve different sizes of fish with broadband sound that cannot be spatially resolved Long-range sonar: Can classify fish through their resonances, discriminate from seafloor and sea surface echoes
Backup
Papers on resonance classification and broadband calibration Stanton, T.K. and D. Chu (2008), Calibration of broadband active systems using a single standard spherical target, J. Acoust. Soc. Am., 124, 128-136. Stanton, T.K., D. Chu, J.M. Jech, and J.D. Irish (2010), New broadband methods for resonance classification and high-resolution imagery of fish with swimbladders using a modified commercial broadband echosounder, ICES J. Mar. Sci., 67: 365-378. Stanton, T.K., C. Sellers, and J.M. Jech (2012), Resonance classification of mixed assemblages of fish with swimbladders using a broadband echosounder at 1-6 khz. Can. J. Fish. Aq. Sci., 69: 854-868. Jones, B.A., T.K. Stanton, J.A. Colosi, R.C. Gauss, J.M. Fialkowski, and J.M. Jech (in manuscript), Broadband classification and statistics of long-range mid-frequency sonar measurements of aggregations of fish, to be submitted to J. Acoust. Soc. Am. (based on Jones 2012 thesis)
Calibration of broadband active acoustic systems using a single standard spherical target Stanton and Chu (2008), JASA
Standard target calibration System response transducer calibration sphere Approach: Normalize measured target response with predicted response Issue: Resonances in scattering response of target can cause error
Source of resonances (shell or solid) Interference in total echo: f FI f (front + c (circumferential interface) waves) + internal waves New approach: use signal processing to isolate f FI completely eliminates resonances
Scattering Amplitude (normalized) Partial Wave Analysis Frequency (khz) ka Standard method: full wave New method: partial wave Impulse Response Analyze echo from front interface only Time (µs)
Largest Calibration Sphere air-filled shell water-filled shell
At Sea Experiment
Results Full wave: contains resonances Partial wave: no resonances Based on partial wave
Target strength, Sonar equation (resolved targets):
Volume scattering strength, Sonar equation (resolved/unresolved targets in volume V):
Net samples Night, upper water column 2008 Atlantic herring Red fish Number of fish Silver hake Butterfish Pearlsides Red hake Length (cm) Squid 0 10 20 30 40
Net samples night, lower water column 2008 Atlantic herring Red fish Number of fish Silver hake Butterfish Pearlsides Red hake Squid 0 10 20 30 40 Length (cm)
Net samples day, upper water column Atlantic herring 2008 Red fish Number of fish Silver hake Butterfish Pearlsides Red hake Length (cm) Squid 0 10 20 30 40
Net samples day, lower water column Atlantic herring 2008 Red fish Number of fish Silver hake Butterfish Pearlsides Red hake Length (cm) Squid 0 10 20 30 40