An ensemble source spectra model for merchant ship-radiated noise

Area of Science:

• Acoustics
• Ocean Engineering
• Signal Processing

Background:

• The classical Ross model for underwater ship noise spectra assumes source spectrum proportionality to a baseline, dependent on ship speed and length.
• Evaluation of the classical model on 54 ship source spectra (30-1200 Hz) revealed significant root-mean-square (rms) errors and overestimated variability.
• Deficiencies stem from weak correlations between source level and ship speed/length in the classical model.

Purpose of the Study:

• To evaluate the classical Ross model for broadband ship noise spectra.
• To propose and evaluate an alternate model to overcome the deficiencies of the classical model.
• To develop a method for generating ensembles of ship spectra.

Main Methods:

• The classical model was evaluated using an ensemble of 54 ship source spectra from 30-1200 Hz.
• An alternate model using a modified rational spectrum with real poles/zeros and non-integer exponents was developed.
• The alternate model's performance was assessed by comparing rms errors against the classical model and frequency dependence analysis.

Main Results:

• The alternate model significantly reduced rms errors compared to models based on a single baseline spectrum.
• A single-term rational spectrum model effectively described high-frequency (400-1200 Hz) ship noise dependence.
• Higher-order models offered only modest rms error reduction at low frequencies (30-400 Hz).

Conclusions:

• The proposed modified rational spectrum model offers a superior approach to modeling underwater ship noise spectra.
• The model's parameters, particularly for high frequencies, provide a more accurate representation of individual ship source levels.
• A joint probability density function for model parameters enables the generation of realistic ship noise spectra ensembles.