Advanced Acoustical Modeling Tools for ESME

1
Advanced Acoustical Modeling Tools for ESME

• Martin Siderius and Michael Porter
• Science Applications Int. Corp.
• 10260 Campus Point Dr., San Diego, CA
• sideriust_at_saic.com
• michael.b.porter_at_saic.com

2
Acoustic Modeling Goals

• Through modeling, try to duplicate sounds heard
  by marine mammals (e.g. SONAR, shipping)
• Develop both high fidelity and very efficient
  simulation tools

3
Acoustic Modeling Goals

• Accurate field predictions in 3 dimensions
• Computational efficiency (i.e. fast run times)
• Propagation ranges up to 200 km
• R-D bathymetry/SSP/seabed with depths 0-5000 m
• Frequency band 0-10 kHz (or higher)
• Moving receiver platform
• Arbitrary waveforms (broadband time-series)
• Directional sources

4
Model Comparisons
Accuracy
Rays
NM and PE
Computation Time
Rays
NM and PE
Frequency
Frequency
5
Motivation
6
Fast Coupled NM Method

• Range dependent environment is treated as series
  of range independent sectors
• Each sector has a set of normal modes
• Modes are projected between sectors allowing for
  transfer of energy between modes (matrix
  multiply)
• Algorithm marches through sectors
• Speeds up in flat bathymetry areas
• Pre-calculation of modes allows for gains in
  run-time (important for 3D calculation)
• Very fast at lower frequencies and shallow water

7
Mid Atlantic Bight Example
8
(No Transcript)
9
Mammal Risk Mitigation Map
SD 50 m SL 230 dB Freq 400 Hz Lat 49.0o
N Long 61.0o W
10
Shipping Simulator

• Using the fast coupled normal-mode routine
  shipping noise can be simulated
• This approach can rapidly produce snapshots of
  acoustic data (quasi-static approximation)
• Self noise can also be simulated (i.e. on a towed
  array)
• Together with a wind noise model this can predict
  the background ambient noise level

11
Example Simulated BTR

• Input environment, array geometry (e.g. towed
  array hydrophone positions) and specify ship
  tracks (SL, ranges, bearings, time)

12
Example BTR from SWELLEX96
13
Computing Time-Series Data for Moving Receiver

1. How is the impulse response interpolated between
   grid points?
2. How are these responses stitched together?

14
1. Interpolating the Impulse Response

• In most cases the broad band impulse response
  cannot be simply interpolated
• For example, take responses from 2 points at
  slightly different ranges

15
2. Stitching the Responses Together

• Even if the impulse response is calculated on a
  fine grid, there can be glitches in the
  time-series data (due to discrete grid points)
• For example, take the received time-series data
  at points 1 m apart

16
Solution Interpolate in Arrival Space

• The arrival amplitudes and delays can be computed
  on a very course grid and since these are well
  behaved, they can be interpolated for positions
  in between.
• Using the exact arrival amplitudes and delays
  at each point, the convolution with the source
  function is always smooth.

17
Ray/Beam Arrival Interpolation
18
Test Case Determine Long Time Series Over RD
Track

• Source frequency is 3500 Hz
• Source depth is 7 m
• Environment taken from ESME test case
• Receiver depth is 7-100 m
• Receiver is moving at 5 knots

19
TL
20
Received Time-Series
21
Received Time-Series
22
Received Time-Series (with Source
Functions)
23
Computing TL Variance

• Fast Coupled Mode approach allows for
• TL computations in 3D (rapid enough to compute
  for several environments)
• Changing source/receiver geometry
• Ray arrivals interpolation allows for Monte-Carlo
  simulations of TL over thousands of bottom types
  to arrive at TL variance

24
Ray/Beam Arrival Interpolation
25
Does it work? TL example

• 100-m shallow water test case
• Source depth 40-m
• Receiver depth 40-m
• Downward refracting sound speed profile
• 350 Hz
• 3 parameters with uncertainty
• Sediment sound speed 1525-1625 m/s
• Sediment attenuation 0.2-0.7 dB/l
• Water depth 99-101 m

26
Does it work? TL example

• Interpolated (red) is about 100X faster than
  calculated (black)

27
TL Variance
28
TL Variance