Abstract
This project’s objective was to improve the quality of acoustic information about marine energy converters that could be collected from groups of drifting hydrophones, while reducing the costs of deployment and data analysis. This was achieved through technology development addressing four focus areas: (1) minimizing flow-noise and self-noise, (2) integrating metadata streams into a single data acquisition system, (3) developing post-processing routines to facilitate rapid data review, and (4) enabling objective identification of marine energy converter sound against a backdrop of ambient noise using time-delay-of-arrival localization. Drifting Acoustic Instrumentation SYstems (DAISYs) consist of a surface expression to monitor position via GPS, a hydrophone recording package at depth, and a compliant suspension system connecting the two. In energetic waves and currents, there are two mechanisms that degrade the fidelity of acoustic measurements: “self-noise” and “flow-noise”. Self-noise is propagating sound produced by an element of the instrumentation system (e.g., metal-on-metal contact for shackles connecting the hydrophone package to the compliant suspension system). Suppressing this is primarily a matter of identifying potential self-noise sources and deploying engineering solutions that mitigate them (e.g., potting shackles in sound-dampening urethane). Flow-noise is non propagating sound of hydrodynamic origin that is produced by acoustic pressure fluctuations around the hydrophone. This is analogous to the “wind in your ears” that one hears when riding a bicycle but is inaudible to a stationary observer nearby. For a drifting hydrophone, flow-noise arises from either relative motion between the hydrophone and surrounding water that causes turbulent eddies to be shed by the hydrophone element or turbulence advected over the hydrophone. For DAISYs in currents, this problem is partially solved when hydrophones drift with the dominant flow, but some residual relative velocity can remain. For DAISYs in waves, the main issue is that the surface expression is forced by the wave field and its motion transferred to the hydrophone package at depth, generating relative motion. In both cases, engineering solutions were identified to suppress relative motion: for currents, a flow-shield creating a quiescent pocket around the hydrophone and, for waves, a mass-spring-damper suspension system analogous to a sonobuoy1. These modifications allow DAISYs to collect data without appreciable flow noise down to frequencies of 10’s of Hz, which are at the lower end of the auditory perception thresholds for marine mammals. Once data are collected, analysis and interpretation are facilitated by semi-automated software to review and compare audio recordings, along with associated metadata streams (e.g., hydrophone depth, spatial position). Software routines developed in MATLAB allow individual DAISY recordings to be simultaneously visualized as spectrograms (acoustic intensity as a function of frequency and time) and listened to as sound. This allows users to build an intuitive understanding of repeated visual elements in a spectrogram (e.g., a high-intensity sound with primary frequency around 1 kHz and duration of < 1 s is the clank of a chain link in a wave energy converter mooring). By identifying differences in the time-of-arrival for the same acoustic event across multiple DAISYs, it is possible to estimate the event’s origin and positively attribute sound to a specific source. An initial DAISY variant – a hydrophone rigidly coupled to the hull of a spar buoy wave measurement drifter – was first used in 2011 to characterize radiated noise around a prototype, scale-model wave energy converter deployed by C-Power in Puget Sound, WA. A similar system was used in 2014 to characterize radiated noise from a prototype river current turbine. At the start of the project in 2017, the DAISY still utilized a version of the wave spar buoy, equipped with a commercial hydrophone package (OceanSonics icListen HF) and augmented by multiple autonomous data loggers to capture metadata describing hydrophone depth, drifter location, ambient metocean conditions (e.g., wind speed), and orientation. This prototype was functional, but faced three challenges: 1. Limited opportunities to reduce cost: The majority of system cost was the commercial hydrophone package and could not be reduced by project team effort. 2. Labor-intensive and error-prone instrument configuration: In the field, individual sensors (i.e., hydrophone, GPS logger, pressure logger, and inertial measurement unit) had to be time-synchronized at the start of each day, offloaded after measurements, and recharged. This meant that DAISYs required several hours of preparation and demobilization each day by a team of trained personnel and it was not uncommon for one or more sensors to be incorrectly configured, resulting in data loss. 3. Flow-noise and self-noise: In currents, propagating sound was often masked by flow-noise at frequencies < 50 Hz and, in waves, flow-noise and self-noise made detecting propagating sound challenging at frequencies < 200 Hz. To overcome these challenges, with support from the team at Pacific Northwest National Laboratory (PNNL), the project evolved the DAISY through three budget periods: • Budget Period 1: Baseline Evaluation (Jan. 2017 – Aug. 2017); • Budget Period 2: System Improvement (Sep. 2018 – Apr. 2020); and • Budget Period 3: Testing in an Energetic Environment (May 2020 – Dec. 2024). The first, brief budget period evaluated the baseline DAISY performance in currents at PNNL’s Marine and Coastal Research Laboratory in Sequim, WA and performance in waves in the adjacent Strait of Juan de Fuca. During the second budget period, which constituted the majority of project activity, DAISYs were upgraded to integrate autonomous sensors and replace the commercial hydrophone package with a less expensive unit and custom-built analog-to-digital converter. Over successive rounds of testing, DAISY reliability improved and issues with flow-noise and self-noise were addressed through flow-shield and suspension system development. The third budget period was also intended to be relatively brief and focus on DAISY deployment around a wave energy converter at the U.S. Navy’s Wave Energy Test Site (WETS). However, unforeseen difficulties delayed any wave energy converter deployment until mid-2024. In the intervening time, the project team made further updates to the DAISY architecture, developed an understanding of flow-noise mechanisms through further field testing in Admiralty Inlet, WA with support through TEAMER, and published a paper describing DAISY development and performance benchmarks. In July 2024, a group of three DAISYs was deployed around C-Power’s SeaRAY at WETS – ironically, a larger version of the scale-model prototype characterized by the first DAISY prototype in 2011. DAISYs were effective at identifying range-dependent noise attributable to the wave energy converter’s power take-off and localizing both wave energy converter noise and odontocete vocalizations. This field deployment was executed by two personnel and the data processing pipeline allowed initial review and characterization of wave energy converter noise by the end of the first day’s deployment. Since project initiation, DAISYs have also been used to characterize noise around multiple current turbines and adapted to a “shallow” version for low-clearance environments that resembles the original wave drifter. Manufacturing plans and processing scripts are freely available through pmec.us/research-projects/daisy. In addition, for those less inclined to build their own instruments, a commercial variant is offered for sale by project partner MarineSitu, Inc. Overall, the project met nearly all its objectives, delivering a step change in our ability to cost-effectively contextualize underwater noise from marine energy converters.