Advanced Solver Options

The Advanced, Parabolic equation and Ray tracer tabs of the Preferences window allow for more detailed control of solver behaviour. Extensive documentation on all types of solvers is available in Reference [2].

Advanced tab

• Autosave: Save the active project file after each solve, if the project has a save file.
• Remove levels below seafloor: dBSeaPE calculates levels in the seafloor as part of a solve. By default, dBSea ignores these levels.
• Levels must decrease with distance from the source: The solver output is often more detailed than required. This option ensures that sound levels decrease monotonically with distance from each source, which is useful when a conservative design is required. This can be applied or removed without recalculating levels in the scenario.
• Stop marching solution upon reaching land (2D PE only): Whenever land is encountered along a slice, the 2D PE solver stops the propagation calculation (on by default). This is useful for limiting computational load next to a large landmass but is not suitable when many small islands are present; when off, the PE continues past thin land features using the seabed properties defined for those cells. Ray solvers terminate rays at land regardless of this setting. The 3D PE solver ignores this setting: the field is set to zero wherever a slice crosses land at each range step, so energy does not pass through land but can couple around it from neighbouring directions. NO_DATA cells in the bathymetry always stop the 2D PE march.
• Frequency oversampling (number of solve freqs per band): dBSea can solve at several frequencies, evenly spaced on a logarithmic axis, within each frequency band (default 1). For example, if the band centre is 1 kHz, the oversampling factor is 5, and the bandwidth is 1 octave, dBSea will solve at 757.9, 870.6, 1000, 1148.7, and 1319.5 Hz, and then power-average the transmission losses to give the level for that band. This option is useful for modelling broadband noise sources and/or fast-changing environments. It increases computation time.
• Radial smoothing factor and exclusion zone radial smoothing factor: The solver output is often more detailed than required. dBSea smooths results or exclusion zones radially around each source, with a triangular kernel of length 2×factor + 1. Defaults are 2 and 1; set a factor to 0 for no smoothing.
• Stop exclusion zone upon reaching land (2D only): Clips the exclusion zone polygon at the coast (on by default).
• Autosave current project after solve: See above.

Minimum attenuation layer thickness, in wavelengths

For dBSeaPE, the computational domain extends below the seafloor and the lowest part of it is given increasing attenuation so that no sound is reflected from the bottom of the domain. Additional attenuation starts at twice the maximum water depth and rises linearly to the bottom of the domain. This setting (default 4) fixes the minimum thickness of that absorbing layer in wavelengths; the layer is never thinner than half the water depth, so the domain is at least 2.5 times the water depth and deeper at low frequencies.

The default value provides good bottom-of-domain absorption for most underwater noise scenarios and is normally fine. If you know that local propagation conditions will drive significant energy into the seabed (for example, strong downward refraction into a soft sediment), consider increasing the sponge layer thickness so that energy reaching the bottom of the domain is fully absorbed rather than reflected back into the water column.

Parabolic equation tab

• Depth and range oversampling: The output from dBSeaPE is calculated at the range points and depth points grid. These oversampling factors (defaults 4 and 0.5, giving a depth step of about a quarter wavelength and a range step of about two wavelengths) control the size of the oversampled calculation grid that dBSeaPE uses internally. Increasing these factors increases the accuracy of the calculated levels at the cost of increased solve time and memory requirements. 3D: angle oversampling (default 1) computes extra azimuthal slices around the source and decimates back to the project’s slices.
• Starting field: dBSea v3 uses Collins’ self-starter for the PE solver, which provides good accuracy across the full frequency range without relying on a separate modal solve for initialisation.

Number of Padé Terms

Specify the number of Padé terms to use (1 to 14, default 6). A higher number will increase the accuracy of the prediction but decrease solving speed; around 5 or 6 terms is sufficient for most scenarios [1]. For the 3D solver, Number of Padé terms in horizontal direction (1, 2 or 4, default 2) controls the azimuthal operator separately.

Broadband: frequency bands per octave

For time-domain PE solves, the PE solver is run at many frequencies spanning the assessment bandwidth, then synthesised into a time series via inverse FFT. The broadband bands per octave setting controls the density of these frequency samples:

Setting	Frequencies per octave	Relative compute time
1/3 octave	3	1×
1/6 octave (default)	6	2×
1/12 octave	12	4×
1/24 octave	24	8×

The default of 1/6 octave is recommended for most scenarios. The acoustic transfer function envelope typically varies slowly enough across frequency that finer spacing does not significantly change the result. Use finer spacing (1/12 or 1/24) only if the environment produces very rapid spectral variations, such as strong modal interference patterns in shallow water.

Ray tracer tab

• Calculate volume attenuation at each step: dBSeaRay can calculate the accumulated seawater absorption along each ray at every step, using the local depth. With this option turned off (the default), absorption is applied once over the straight-line path length using the coefficient at the mean of the source and receiver depths. The volume attenuation is typically negligible except at high frequencies.
• Collect ray arrivals at probes: Stores the arrival list at each probe for export (off by default).
• Number of rays per slice, Initial angles range: dBSea splits the sound leaving the source into N evenly spaced rays in declination between the start and end angles (default −89° to +89°; 0 degrees is horizontal from the source, -90 degrees is up, and +90 degrees is down). Leave the count at 0 for the default of 5000. In 3D, Number of ray divisions horizontally sets the number of azimuths (0 for the default of 1000).
• Max number of reflections from seafloor (per ray): After this number of reflections, the ray is terminated (default 100). In shallow, long-range scenarios rays may need many more bounces to reach the far field, so increase it if levels fall off unexpectedly.
• Initial step size (m): Size of the first step in metres (default 5). Following step sizes depend on solver grid resolution and ray path.

References

[1] M.D. Collins, Higher-order parabolic approximations for accurate and stable elastic parabolic equations with application to interface wave propagation. J. Acoust. Soc. Am. 89, 1050–1057 (1991).

[2] F.B. Jensen, W.A. Kuperman, M.B. Porter, & H. Schmidt, Computational Ocean Acoustics, 2nd edition, Springer (2011).