2025 CSDMS meeting-115
From CSDMS
(if you haven't already)
Log in (or create account for non-CSDMS members)
Forgot username? Search or email:CSDMSweb@colorado.edu
Browse abstracts
Leveraging FUNWAVE-TVD Outputs as Training Data for Nearshore ML Models
Ryan Schanta,
University of Delaware Newark Delaware, United States. rschanta@udel.edu
Wave-resolving, Boussinesq nearshore wave models such as FUNWAVE-TVD are capable of providing nonlinear hydrodynamic outputs that wave-averaged models cannot directly provide. Understanding such nonlinear nearshore processes is crucial to deepen our understanding of complex coastal processes, such as morphodynamics and sediment transport. However, the computational cost of wave-resolving models has made them prohibitive to use for many such applications.
To bridge this gap, a machine learning model trained on thousands of FUNWAVE-TVD models using synthetic, experimental, and field data was developed to estimate nonlinear nearshore wave statistics. Given boundary conditions and forcing terms, the model can “learn” the statistics associated with nonlinear nearshore processes. Such a model is broadly useful for other coastal models that rely on accurate measures of these nonlinear wave properties to parameterize other processes of interest.
