Short waves are of key importance for nearshore dynamics, particularly under storms, where they contribute to extreme water levels and drive large morphological changes. Therefore, it is crucial to model accurately the propagation and dissipation of storm waves in the nearshore area. In this paper, field observations collected in contrasted environments and conditions are combined with predictions from a third-generation spectral wave model to evaluate four formulations of wave energy dissipation by depth-induced breaking. The results reveal a substantial over-dissipation of incident wave energy occurring over the continental shelf, resulting in a negative bias on significant wave height reaching up to 50%. To overcome this problem, a breaking coefficient dependent of the local bottom slope is introduced within depth-induced breaking models in order to account for the varying degrees of saturation naturally found in breaking and broken waves. This approach strongly reduces the negative bias observed in the shoreface compared to default parameterizations, yielding significant improvements in the prediction of storm waves. Among the implications of this study, our new parameterization of the breaking coefficient results in systematically increased predictions of the wave setup near the shoreline compared to the default parameterization. This increase reaches a factor 2 for gently sloping beaches.