Characterization and modeling source levels of commercial vessels in the Gulf of Mexico.

The Gulf of Mexico is among the noisiest marine regions globally, primarily due to widespread seismic airgun operations and vessel traffic. While airguns dominate the low-frequency soundscape, vessel traffic also contributes substantial high-amplitude noise in the same range low-frequency band (<500 Hz). Between August 2020 and July 2022, two underwater acoustic recording stations documented 13 930 vessel transits from five major ship types operating within commercial shipping lanes. Tankers and cargo ships were the most common, followed by tug-tows, passenger ships, and special crafts. Cargo ships and tankers had average broadband (20-1000 Hz) monopole source levels (MSLs) of ∼183 dB re 1 μPa m, while tug-tows were 2-3 dB lower, and passenger ships/special craft were 4-5 dB lower. To investigate factors influencing low-frequency sound production, this study analyzed the relationship between vessel MSLs and ship characteristics, transit conditions, and oceanographic parameters. For this study, machine-learning models were trained to predict MSLs and their performance was compared to that of generalized additive models. Vessel speed was the most influential predictor, with additional contributions from deadweight, gross tonnage, length, and environmental parameters. This machine learning approach provides a tool to estimate MSLs in other regions and simulate the effects of noise reduction solutions, such as speed reduction or vessel design modifications.