Author(s): Andy Plater; Jonathan Higham; Ben Phillips; Nicoletta Leonardi; Dani Arribas-Bel; Cai Bird

Linked Author(s): Jonathan Higham, Nicoletta Leonardi

Keywords: Computational methods; Coastal processes; Hydrodynamics; Sedimentation; Bathymetry

Abstract: To fully examine the underlying processes of coastal change we need to determine the largest underlying driving mechanisms (hydrodynamics and geomorphological change). The associated mechanics of these mechanisms are based on a vast spectrum of spatial and temporal scales, all of which are important. While independent methods can monitor and model these different scales at different locations obtaining time-synchronized real-world view across all these scales is beyond the scope of current technologies. In this study, we present a novel method and early findings of our machine learning, Eigen decomposition-based data assimilation technique “Eigenshores”. Using this method, we assimilate spatially and temporally different datasets. Applying these machine learning-based techniques we create a probabilistic view of short-term coastal changes i.e. with good accuracy we can forecast changes in hydrodynamics and geomorphic change at a vessel level scale creating an early warning system for low carbon and safe port navigation.

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221480

Year: 2022