to be retrained on relatively small regional datasets, substantially reducing data demands. Synthetic data can further enhance results by blending animal signals with local background noise to represent new environments. AI can dramatically accelerate analysis, but its value depends on the systems around it. Models are only as effective as the data pipelines, validation processes and organisational structures that support them. As marine science becomes more automated and data-intensive, building the right tools and workflows is just as important as deploying the right algorithms. Designing systems that scale Innovation in marine science isn’t only about new sensors or smarter algorithms. Much of the progress comes from having the right tools, workflows and organisational systems to allow these technologies to scale. Underwater noise modelling is a clear example. GHD research has found that many commercial software packages lack transparency, struggle with regulatory requirements or cannot easily handle large data volumes. 11 To address this, our teams developed an in-house acoustics modelling toolbox using the data analysis-focused programming language R, designed to improve repeatability and make assumptions explicit. Scripted workflows also allow us to integrate commercial tools where useful while maintaining full control over data processing, visualisation and reporting. Custom tooling also supports advanced modelling needs. For a project commissioned by the Norwegian government, researchers developed bespoke scripts to analyse fish movement and behavioural responses to seismic noise. 12, 13 Linking acoustics, movement data and statistical modelling would not have been possible using packaged solutions alone. Emerging technologies will further expand these
capabilities. Autonomous underwater vehicles (AUVs) are already filling the gap between ship-based surveys and stationary monitoring systems. Swarm robotics is beginning to show potential for large-scale monitoring tasks. 14 Next-generation animal-borne tags are becoming smaller and more capable, with “grain-of-rice” sensors now collecting depth, temperature and acceleration data for up to 40 days, and transmitting across hundreds of metres. 15 Tools only deliver value when organisations know how to use them. Cross-regional knowledge hubs and communities of practice, like GHD’s 50-member marine science network, are essential for sharing methods, refining workflows and ensuring consistent delivery across diverse regulatory settings. A hybrid future for marine science Marine science is entering a period defined by scale: more activity offshore, more environmental commitments and more data than ever before. In response to this challenge, the most meaningful innovations are emerging not from just chasing the newest tool, but from connecting existing and emerging technologies. Digital aerial surveys, PAM systems, eDNA sampling, advanced modelling and machine learning all contribute different pieces of insight. Together, they form faster, clearer and more defensible evidence for complex marine decisions. As governments pursue global biodiversity targets and offshore development accelerates, this hybrid model will become increasingly essential. Organisations that can integrate methods, share knowledge and apply technology pragmatically will be well-placed to lead. For marine science, the future is not just digital or automated; it is connected.
10 | GHD | Nexus Magazine
Made with FlippingBook - professional solution for displaying marketing and sales documents online