30 MARCH 2022 • WORLD AQUACULTURE • WWW.WA S .ORG data. For local observation, aquaculture sites may be fitted with audio or visual recording devices, such as video cameras or hydrophones. Data is typically stored onboard in these devices that are periodically physically collected at the site and returned to shore for analysis. Processing of data consists of tedious manual observation of video or audio data to identify any signs of species of interest. Such activities are obviously resource intensive and, while they may help to characterize animal activity in the vicinity of an offshore facility, they do not provide an opportunity for intervention if an entanglement threat is observed. Next-Generation Marine Mammal Monitoring There is, then, a need in the current monitoring landscape for tools that are both automated, to reduce manpower needs of wildlife observation, and real-time, and therefore able to prompt timely intervention and rescue efforts. The evolution of two major technological focus areas – machine learning and the Internet of Things (IoT) – help bring these needs to reality. Machine learning uses large amounts of annotated (i.e., preprocessed) data to train software algorithms to perform tasks previously reserved for human operators and can automate the interpretation of video and audio data collected onsite, in particular the detection and classification of species of interest. The Internet of Things refers to a range of technologies facilitating deployment of embedded sensors and processors and the ability of these devices to communicate with one another and through larger networks. Application of IoT technologies to monitoring systems allows real-time communication between offshore unmanned systems and onshore users via cellular networks, satellites or other wireless communication protocols. Research and product development are currently underway to harness these technologies to address the challenge of marine entanglement. The DeepSeaVision-AI system, developed by Synthetik Applied Technologies as part of a NOAA-funded research program, provides an integrated hardware and software platform for the real-time automated detection of marine mammals at offshore facilities. The integrated hardware platform consists of embedded processors, communication hardware and visual and audio sensors. Collected data is processed onboard the device to Expansion into offshore waters has become an important issue in the aquaculture industry, as the global blue economy seeks to expand seafood production. Although offshore aquaculture has significant potential to increase total capacity (FAO 2013), the concept of shifting aquaculture further from the coastline has brought with it several environmental concerns, which in turn has created a challenging permitting landscape. One of the chief concerns with offshore aquaculture in the United States is the potential entanglement risk for protected species with fishing gear and similar obstructions in the water. The entanglement issue is most starkly represented by plight of the North Atlantic right whale – a species with fewer than 400 individuals remaining in the wild – with the majority of recorded deaths since 2017 attributed to gear entanglement or vessel strikes (NOAA Fisheries 2021). Outside of the North Atlantic Ocean, different species drive entanglement concerns in different regions, such as the Bryde’s whale population in the Gulf of Mexico or monk seal populations around Hawaii. Concerns with entanglement have helped contribute to a permitting landscape for offshore aquaculture in the United States that has become intractable. The process requires authorizations from a plethora of federal agencies with no single definition of what constitutes acceptable mitigation of entanglement risk, further complicating matters. To fully bring the promise of offshore aquaculture to bear, tools to help facility operators effectively prevent and mitigate entanglement events and reassure federal stakeholders are critical. Many technological advancements are being explored to reduce the likelihood and impact of entanglement, such as ropeless systems or gear that will break away under tension loads. Alongside these physical measures, more effective monitoring systems can help increase understanding of population dynamics and animal behavior to better characterize entanglement risk and alert operators of imminent entanglement events. Unfortunately, current solutions for monitoring protected species near offshore facilities can be time-intensive and critically do not provide real-time information that can be used to trigger rescue or other mitigation measures. Shipboard or aerial surveys can help to understand population distributions but do not help researchers understand how species interact with aquaculture facilities and only provide a snapshot in time rather than continuous Acoustic Detection of Marine Mammals for Entanglement Prevention Josh Hatfield FIGURE 1. An Internet of Things-based monitoring system deployed at the University of New Hampshire Aquafort platform. Video and audio data are processed using machine learning models aboard the device.
RkJQdWJsaXNoZXIy MjExNDY=