Application of a data-model assimilation method to a 3D surf zone sandbar physical experiment
Résumé
Video imagery recently arose as a low-cost alternative to direct measurement to estimate wave characteristics and then tackle the lack of actual updated bathymetries hydro-morphodynamic coupled numerical models typically face. Nonlinearities that are ubiquitous over complex three-dimensional surfzone sandbars are a key problem for observation relevance, and feed an increasing need of local wave information. Therefore, a close interaction between video imagery and modeling is highly required through observation assimilation. In this paper, we explore two bathymetry inversion methods related to an original experiment on nature-like beaches. Wave celerity was first video-estimated and depth inversion was performed to quantify the underlying bathymetry. While existing methods were previously tested over rather alongshore-uniform barred beaches, here we show that the efficiency of such method strongly decreases with increasing beach alongshore non-uniformity. A data-model assimilation, based on a Kalman filter, was further developed and applied to our experimental study using bathymetric data that derived from wave celerity estimates. Analytically set Gaussian-type covariance matrices were inverted to provide local correction. The method shows promising skills to correct bathymetry with the single source of bathymetric video proxy and now required further in-depth improvement.