Experimental validation of non-linear deterministic prediction schemes for long-crested waves

Abstract

This paper describes the experimental validation of non-linear deterministic prediction tools for unidirectional seas, based on the use of two different non-linear wave models: (i) a so called “second order model enhanced”, which is analytic, and (ii) the numerical HOS wave model, which is run at order 3 in our simulations. Both prediction models use time traces of the surface elevation measured at one or several wave probes as an initial condition. Two effective data assimilation processes are used to optimize the model-reconstructed probes measurements, and thus ensure proper initial conditions for the forecast. The validation of our prediction tools on model-generated wave data has been reported in the previous paper (Blondel et al., 2010). Accurate predictions were reported compared to linear predictions. In the present study, the predictions are compared to different sets of laboratory measurements collected in a towing tank. The good agreement between the model-predicted surface elevations and the related experimental measurements indicates that our models can predict irregular long-crested seas more accurately than basic linear approaches with computational costs compatible with the short-term constraint.