Marine Soundscapes - discriminating and understanding biological and anthropogenic sources of sound from autonomous vehicles

声音是海洋哺乳动物和许多鱼类交流、搜索猎物和导航以避开捕食者和危险的核心。人为噪声源可能掩盖这些具有生物重要性的声音，并可能造成可能延伸至海洋环境大片区域的生理和行为影响。海洋勘探的新前沿领域之一是使用海洋自动驾驶船队来收集海洋数据。电池效率和计算机技术的改进意味着被动声学监测设备现在可以安装在这些交通工具上，从而能够以前所未有的时间和空间分辨率确定海洋中的声景。这些PAM设备记录了大量数据，需要新的信号处理方法和大数据方法来实现有效的实时数据传输和任务后分析。因此，该项目将开发机器学习技术，利用从海上自动驾驶车辆船队收集的PAM数据，区分不同的声源。除了提高我们对海洋中声音的了解外，该项目开发的技术还与监测自然海底渗漏以及工业基础设施(管道和钻井平台)以及碳捕获和储存项目的泄漏直接相关。

Sound is central to the way marine mammals and many fish species communicate, search for prey, and navigate to avoid predators and hazards. Anthropogenic noise sources may mask these biologically important sounds and can cause physiological and behavioural impacts that may extend over large areas of the marine environment. One of the new frontier areas of ocean exploration is using fleets of marine autonomous vehicles to collect oceanographic data. Improvements in battery efficiency and computer technology mean that Passive Acoustic Monitoring (PAM) devices can now be carried on these vehicles enabling soundscapes within the oceans to be determined at unprecedented temporal and spatial resolution. These PAM devices record huge volumes of data that require new signal processing methodologies and big data approaches to enable effective real-time data transfer and post-mission analysis. This project will therefore develop machine-learning techniques to discriminate between different sources of sound, using PAM data collected from fleets of marine autonomous vehicles. As well as improving our understanding of sound in the oceans, the techniques developed in this project are directly relevant to monitoring of natural seabed seeps as well as leaks from industrial infrastructure (pipelines and rigs), and from carbon capture and storage projects.