This thesis will build a physics-based model to predict underwater noise levels for frequencies between 0.5-20 kHz. These mid-frequencies are important for SONAR applications. The noise in this bandwidth is mostly due to wind. The wind blows over the ocean to create waves, which then break to form bubbles of different radii. The total number of bubbles potentially reach up to an order of 106 to 108 per m3. These bubbles that are formed due to this air-sea interaction, and entrapped in the water column, are the main sources of noise.



Most of the previous works have used empirical noise models based on measurements in the ocean. All these empirical noise models provide noise spectra as a function of wind speed or sea state. There are mainly two reasons for the development of a physics-based model rather than only an empirical one. The first is that empirical models for noise levels use the sea-state, a subjective quantity based on visual observations, which may not always be available. The second is that observations have always differed from these predictions at high frequencies (>8 kHz), and fast wind speeds (>10 m/s). This thesis will go beyond the existing models by using recent literature on the formation of bubbles that suggest occurrence of bubbles in cloud-like formations, which evolve across space and time. There are yet no quantitative underwater noise models that combine the nature of these clouds, with their acoustics. This thesis will use the spatial and temporal scales of these bubble clouds to construct an acoustic prediction model at high frequencies. By building a physics-based noise model for bubbles, the thesis will also explain the anomaly at high wind speeds.

This thesis will build the model in two steps. First, inputs such as wind speed and wave height will be used to approximate the clouds as spheres, with respective number of bubbles of each radius and mean spacing between the clouds. Second, this approximate model will be used to predict the radiated sound. To test the model, predictions will be setup for comparison with measured noise during high winds around Monterey, California. To input relevant environment parameters to the model, the work will use measurements of wind-velocities, and wave-heights, measured on-site as well as from re-analysed wind data from ECMWF. For acoustic recordings, the work will use hydrophones from the Monterey Accelerated Real Money Rummy System (MARS) cabled observatory and High Frequency Acoustic Recording Package (HARP).