Abstract
The Offshore Energy Research Association of Nova Scotia (OERA), the Technology Strategy Board of the United Kingdom (now Innovate UK) and the Province of Nova Scotia signed a Memorandum of Understanding (MOU) providing the framework to establish bilateral collaboration in tidal research between Canada and the United Kingdom. The aim of the MOU was to fund innovative industrial research to address current and compelling knowledge gaps related to sensing and monitoring technologies with the goal of improving quality and quantity of data available to the industry. The collection and processing of data relating to marine life in tidal energy sites has so far been limited to individual use of either active acoustic monitoring (AAM) or passive acoustic monitoring (PAM). A project team was assembled for this project to explore the amalgamation of AAM (sonar) and PAM (hydrophone) data to facilitate the development of improved sonar detection and classification software. The project is titled Integrated Active and Passive Acoustic System for Environmental Monitoring of Fish and Marine Mammals in Tidal Energy Sites (ISEM).
This integrated monitoring system incorporated two different sensors: the Tritech Gemini Imaging Sonar (active acoustic) and the Ocean Sonics icListenHF smart hydrophone (passive acoustic). These sensors were co-located on an OpenHydro turbine subsea structure and the team explored the creation of a data interface that would allow data from each sensor to be combined into an integrated fish and marine mammal data set.
The initial ISEM project objectives were as follows
- Improve existing sensor technology software to maximize individual sensor capability;
- Integrate two complementary sensor technologies to improve ability to detect, classify, localize and track marine mammals (notably harbour porpoises) and fish in real-time; and
- Test sensor capabilities and integrated system effectiveness in high energy sites in the Bay of Fundy.
Berth D, at Fundy Ocean Research Center for Energy (FORCE), was the location for the Cape Sharp Tidal (CST) project, a joint venture research project between OpenHydro, a Naval Energies company, and minority partner Emera Inc. involving the deployment of an Open Centre design gravity-base instream tidal turbine. ISEM devices were installed on the instream tidal turbine for a first deployment that took place in November 2016 with the turbine retrieved in June 2017. A second deployment took place on 22 July 2018; the turbine and all associated devices were isolated from the power grid following the announcement of OpenHydro’s insolvency on 26 July 2018.
Due to the setbacks experienced as part of the CST project, the ISEM project objectives were adjusted from the original objectives laid out. Amalgamation of AAM and PAM data sets was not achieved, which limited the development of improved sonar detection and classification software, but a number of advancements were made in understanding the utility and challenges within each of the data streams to allow for future integration.
Passive Acoustic Monitoring
Four icListen hydrophones were installed on the turbine infrastructure in order to provide all-around porpoise (and dolphin) detection coverage of the turbine and theoretical localization of inbound echolocating porpoises at distances beyond the near-field. However, a number of issues arose which led to hydrophone performance and localization challenges, as summarized below:
- Loss of communication related to cabling;
- Sampling rates of .wav data sets during the first deployment were insufficient for full analysis using new porpoise detection software;
- Hydrophone damage from debris travelling in the water column;
- Masking from high background noise in the environment from tidal currents, turbine operation and signals emitted from various sensors;
- Large separation distance between individual hydrophones resulted in few vocalizations occurring on multiple hydrophones; and
- Turbine structure obstruction (further reduction in ability to achieve concurrent vocalization detections).
These challenges led to recommendations and lessons learned for future marine mammal monitoring projects. Some of the main recommendations are:
- Cables to be inspected and tested prior to deployment to ensure integrity, and sufficient data transfer rates when transferring data from multiple instruments simultaneously;
- Reinforced guards to be used to protect sensor elements (improved guards led to improved data during second turbine deployment);
- An analysis of the required positioning and spacing between hydrophone units for optimal localization of porpoise clicks and consideration for multiple hydrophone array/clusters;
- Placement of hydrophones to limit impact of noise emitted from other equipment; and
- Post-deployment testing to measure the sound profile of each instrument on the turbine.
Notwithstanding the above challenges, the Coda porpoise click detector, developed independently of the project for Ocean Sonics Ltd, performed well in datasets with strong tidal current noise. Further processing with the Coda+ program can use the click results to detect porpoise click trains and create probability models. During the second turbine deployment, Coda was integrated into an automated program in order to keep the full bandwidth .wav files with detected porpoise clicks in real time. The automated program was not able to be fully tested or optimized due to the shutdown of the monitoring equipment following the announcement of OpenHydro’s insolvency on 26 July 2018. Although there was insufficient time to run complete testing of the detector, Coda was shown to be functional on the substation computer before the deployment and performed well during the two days of monitoring post-deployment.
Given Coda’s real-time analysis of .wav data to down-sample data and produce diagnostic output in text form, Coda shows the potential for interfacing with the software controlling other devices. PAMGuard’s real-time diagnostic outputs has similar potential but would require a software update to be compatible with icListen input data. Integrating the automatic detection of clicks with an Active Acoustic Monitoring (AAM) device could serve to increase the reliability of detection, localization and potential interactions of porpoise around a tidal turbine.
Active Acoustic Monitoring
The use of the Tritech Gemini sonar to manually observe and record large individual fish, schools of fish and other sea life (marine mammals, sharks) is now well established. Turbulence imparted by tidal flows or turbulent wakes, however, was shown to degrade the Gemini image quality, and thus limit fish target detection.
SeaTec software was used to automatically observe and record schools of fish in this project, albeit with mixed results. The main problem is one of recording false positives than missing false negatives. The reasons for this can be summarized as:
- Too much noise in defined localities of the image (e.g. backscatter from the seabed);
- Recording tidal/drift targets as sea life;
- Incorrect identification of flickering but non-moving targets; and
- Recording different parts of the same target as multiple targets.
Steps taken to correct these issues and implemented in the SeaTec software include:
- Use of configurable exclusion zones;
- Allow input of tidal direction (manually or via serial port) to filter out inanimate drifting targets;
- Evaluation and exclusion of static targets; and
- Use of “group” classification (especially useful for large targets like sharks).
Tritech also undertook software engineering to allow additional data streams (e.g. from hydrophone detections) to be incorporated within the SeaTec software architecture.
A short validation of marine mammal targets identified by the SeaTec software was undertaken. Similar to fish detection, false positives were often noted, mainly determined to be reflections from surface waves. The use of the SMRU marine mammal Classifier (using movement and shape parameters of validated targets) was helpful to identify higher probability targets and as such can help in efficiencies involved in the human validation process and removal of false positives. Nevertheless, a human validation step is still required after detection by the SeaTec software.
Overall, ISEM project results show that there is potential to use integrated active and passive acoustics to monitor harbour porpoise and tagged fish in the near field of tidal turbines. Coda was able to detect harbour porpoise clicks while the Gemini was operational. Localization of near-field porpoises, however, will likely require additional hydrophones.
Lessons learned and recommendations are highlighted in Sections 6.0 & 7.0, respectively, and include the need for frequent and direct communication between tidal turbine developers and both researchers and sensor/software developers to ensure essential sensor testing prior to deployment, proper equipment setup, monitoring of equipment, and data management. In particular, it is vital that engineers installing equipment provide verification of device settings and device alignment prior to deployment, as well as feedback on input data as soon as data is received after deployment.
Both PAM and AAM datasets result in high volumes of data (10-15 TB per month in this study) and these cannot be easily transferred or viewed without appropriate protocols and technology. A data management plan is essential for project success. It needs to be developed in advance of a project to ensure high quality data collection, long-term data storage, and timely access to the data for processing and analysis.