Abstract
Turbulence-induced pseudosound, or flow noise, is a ubiquitous challenge encountered when making underwater acoustic measurements in energetic environments, such as tidal channels or rivers. Flow noise is observed at relatively low frequencies – though both the extent of affected frequencies and the magnitude of flow noise increase with current speeds. In strong currents (i.e., above 3 m/s), flow noise has been observed at frequencies above 500 Hz, complicating the characterization of low-frequency sound emissions from current energy converters. The magnitude of flow noise is also dependent on the dimensions of the recording transducer: turbulent length scales larger than the longest dimension of the transducer will engulf the transducer, resulting in the highest levels of flow noise, while length scales smaller than the transducer will either completely or partially cancel around the body of the transducer.
Flow noise is relatively straightforward to identify in acoustic recordings because of its characteristic decrease in intensity with increasing frequency due to the roll-off in turbulent kinetic energy. While this makes it easy to identify and isolate flow-noise contamination from data, flow noise may still mask low-frequency sounds of interest. As such, flow noise mitigation through mechanical design is desirable. A variety of flow noise mitigation approaches have been used in the literature with varying success, but few have been quantitatively evaluated.
Here, we aim to quantitatively assess two distinct approaches to flow noise mitigation: straightening the flow around the hydrophone transducer and increasing the effective diameter of the transducer. The first approach involves a shroud or housing around the hydrophone designed to straighten (i.e., reduce turbulence in) the flow around the hydrophone. In this work, we test three such designs: an open cell foam housing, a thin nylon shroud, and a ballistic nylon shroud. The second approach increases the effective dimensions of the hydrophone transducer in the direction of the flow, which increases the length scales of turbulent pressure fluctuations that contribute most significantly to flow noise. In this work, we test this approach using an oil-filled polyurethane covering that effectively increases the largest dimension of the hydrophone transducer by approximately a factor of four.
The four flow shield designs are evaluated in a narrow tidal channel in Sequim Bay, WA. Each flow shield is deployed on a hydrophone alongside a second unshielded hydrophone for one tidal cycle. Concurrent flow measurements are made with an acoustic doppler velocimeter (ADV) to provide contextual information about flow speeds and turbulence, and an acoustic projector deployed nearby is used to assess attenuation associated with the flow shield. Results from each design are presented, including comparison of flow noise and received levels between the shielded and unshielded hydrophones. We also make recommendations for future testing and design refinement.