Periodic Reporting for period 1 - SEASOUNDS (Innovative marine soundscape characterization to effectively mitigate ocean and sea noise pollution)
Período documentado: 2024-01-01 hasta 2025-12-31
SEASOUNDS aims to better characterize and predict marine soundscapes in various European eco-regions (e.g. parts of the Mediterranean and the Baltic Seas, the Venice Lagoon, the Norwegian fjords) where the scarcity of scientific data, upon which decisions are based, as well as the difficulties associated with sound field monitoring, processing and understanding, make underwater noise pollution poorly known and hence insufficiently regulated. One of the main goals of the project is to provide recommendations for appropriate and proportionate underwater noise mitigation solutions, for improved know-how, decision-making and standards setting for a sustainable Blue Growth limiting the impact of noise pollution on marine wildlife. SEASOUNDS aims to have a strong impact on society also by raising awareness among the general public, and more specifically young public, on a form of pollution often underrated or overlooked.
SEASOUNDS addresses important knowledge gaps related to understanding, characterization and modeling of the entire noise transfer chain, from the noise source (e.g. offshore foundation installation, disposal of unexploded ordnance, shipping) to the receiver (whether a technological tool or an animal such as marine mammal, fish and invertebrate). Highly multidisciplinary, SEASOUNDS' methodological approach is built around the idea that, to address effectively these complex scientific gaps, we need to go beyond the classical underwater acoustics-related approaches and incorporate concepts, models, and tools from various scientific fields such as underwater acoustics, seismology, mechanics, bioacoustics, ecoacoustics and marine biology. The project also takes advantage of cutting-edge technologies developed in these fields, like optical cables deployed on the seabed allowing Distributed Acoustic Sensing (DAS) data processing techniques based on Artificial Intelligence (AI), and High-Performance Computing (HPC).
SEASOUNDS addresses important knowledge gaps related to understanding, characterization and modeling of the entire noise transfer chain, from the noise source (e.g. offshore foundation installation, disposal of unexploded ordnance, shipping) to the receiver (whether a technological tool or an animal such as marine mammal, fish and invertebrate). Highly multidisciplinary, SEASOUNDS' methodological approach is built around the idea that, to address effectively these complex scientific gaps, we need to go beyond the classical underwater acoustics-related approaches and incorporate concepts, models, and tools from various scientific fields such as underwater acoustics, seismology, mechanics, bioacoustics, ecoacoustics and marine biology. The project also takes advantage of cutting-edge technologies developed in these fields, like optical cables deployed on the seabed allowing Distributed Acoustic Sensing (DAS) data processing techniques based on Artificial Intelligence (AI), and High-Performance Computing (HPC).
During the reporting period, the project achieved significant progress across its technical and scientific axes by integrating numerical modelling, data-driven signal analysis, field data acquisition, and experimental development. Foundational knowledge was established through comprehensive literature reviews covering predictive modelling for impact and vibratory piling, underwater acoustic propagation, mitigation strategies, and biological impacts on benthic organisms. In the domain of Distributed Acoustic Sensing (DAS), signal-processing workflows were developed to compute power spectral densities and frequency–wavenumber (FK) spectra (Deliverable D3.1). These tools facilitate seafloor wave-propagation analysis, while investigations into fibre burial depth clarified its impact on cable–sediment coupling. A uniquely valuable six-week multi-sensor dataset comprising DAS, Ocean Bottom Seismometer (OBS), and hydrophone data was collected in a fjord environment. Machine-learning efforts produced a convolutional autoencoder to cluster compressed signal representations and a shared toolbox for large-scale data annotation.
For UXO detonation modelling, semi-analytical simulations using the OASES code quantified the strong influence of sediment properties—such as layer thickness and shear-wave velocity—on seismo-acoustic propagation. A hybrid modelling path was established using OpenRadioss to derive equivalent source models for non-linear detonation signatures, serving as inputs for 3D spectral-element simulations in SPECFEM. Adjoint-based sensitivity kernels for pressure, particle velocity, and sound exposure levels were successfully derived, providing a rigorous mathematical framework for uncertainty quantification. Experimental progress centered on the AquaVib system, where a fluid-structure interaction (FSI) model was developed to disentangle acoustic pressure (AP) and particle motion (PM). Bioacoustic modelling reached a milestone with the implementation and validation of a high-resolution pseudospectral protocol for toothed-whale hearing anatomy, resulting in a peer-reviewed publication in the Journal of the Acoustical Society of America. Ecological studies advanced for seal drift dives and Venetian lagoon soundscapes.
For UXO detonation modelling, semi-analytical simulations using the OASES code quantified the strong influence of sediment properties—such as layer thickness and shear-wave velocity—on seismo-acoustic propagation. A hybrid modelling path was established using OpenRadioss to derive equivalent source models for non-linear detonation signatures, serving as inputs for 3D spectral-element simulations in SPECFEM. Adjoint-based sensitivity kernels for pressure, particle velocity, and sound exposure levels were successfully derived, providing a rigorous mathematical framework for uncertainty quantification. Experimental progress centered on the AquaVib system, where a fluid-structure interaction (FSI) model was developed to disentangle acoustic pressure (AP) and particle motion (PM). Bioacoustic modelling reached a milestone with the implementation and validation of a high-resolution pseudospectral protocol for toothed-whale hearing anatomy, resulting in a peer-reviewed publication in the Journal of the Acoustical Society of America. Ecological studies advanced for seal drift dives and Venetian lagoon soundscapes.
The project has moved significantly beyond the state of the art in several key areas. The non-linear FSI model of the AquaVib system is the first to resolve membrane-driven AP and PM fields simultaneously, enabling dose-response experiments where particle motion—a critical stimulus for marine invertebrates—can be isolated. In odontocete bioacoustics, the pseudospectral protocol applied to realistic anatomy translates raw environmental noise into indicators of communication masking and localisation performance, offering a grounded pathway to address chronic sub-lethal effects that current regulatory frameworks struggle to quantify.
The adaptation of adjoint sensitivity analysis from seismology to underwater noise introduces quantitative tools for propagating environmental uncertainty into regulatory indicators. The UXO source model overcomes traditional limitations by coupling multiphase detonation signatures with full-wave 3D propagation. Furthermore, the integration of systematic coupling studies with machine learning and open-source toolboxes creates a scalable pathway for operational DAS monitoring, moving beyond single-site case studies toward reproducible workflows. These results improve scientific credibility, biological relevance, and cost-effectiveness. Future success requires field validation of models, sustained access to high-performance computing, and the harmonisation of measurement conventions for particle motion. Software sustainability and open data sharing will ensure technical advances reach international researchers and industry stakeholders.
The adaptation of adjoint sensitivity analysis from seismology to underwater noise introduces quantitative tools for propagating environmental uncertainty into regulatory indicators. The UXO source model overcomes traditional limitations by coupling multiphase detonation signatures with full-wave 3D propagation. Furthermore, the integration of systematic coupling studies with machine learning and open-source toolboxes creates a scalable pathway for operational DAS monitoring, moving beyond single-site case studies toward reproducible workflows. These results improve scientific credibility, biological relevance, and cost-effectiveness. Future success requires field validation of models, sustained access to high-performance computing, and the harmonisation of measurement conventions for particle motion. Software sustainability and open data sharing will ensure technical advances reach international researchers and industry stakeholders.