Transforming metocean data into motion predictions: A review of very short-term predictions of floating body motions

This paper introduces a novel, phase-based classification framework for the very short-term prediction (VSTP) of floating body motions to resolve the ambiguities and overlaps inherent in traditional algorithm-based classifications. The proposed framework organizes existing approaches into three forecasting phases, i.e., wave-field measurement, wave-field forecasting, and wave-to-motion mapping. It identifies four major model categories: wave forecasting, hydrodynamic, adaptive filtering, and historical data–driven models. Notably, this structure clarifies the functional roles of different methodologies and resolves long-standing classification ambiguities. Furthermore, a bibliometric analysis of the literature from 2015–2025 demonstrates that research activity has increasingly shifted toward data-centric approaches. Indeed, publications on historical data–driven models increased from 8 in 2015 to 39 in 2024, corresponding to a compound annual growth rate (CAGR) of approximately 20%, while adaptive filtering models exhibited even faster growth, rising from 5 to 32 studies with a CAGR of about 25%. In contrast, wave forecasting models showed moderate growth, increasing from 2 to a peak of 15 before declining to 11, whereas hydrodynamic model studies remained relatively stable at 0–7 publications per year. Comparative performance analysis reveals that no single model consistently outperforms others across key criteria, including prediction accuracy, forecasting horizon, interpretability, and real-time applicability. Instead, future advances in VSTP are likely to rely on multi-model fusion, improved wave-sensing technologies, data augmentation for extreme sea states, and uncertainty-aware forecasting, all of which are essential for reliable deployment in safety-critical offshore operations.