Currently, morphodynamic models as XBeach show substantial overestimations of the erosion rates during breaching and overwash events at barrier islands. The presently used limitations on the Shields parameter and the sediment concentration do hinder erosion, but have undesirable side effects, e.g. the breaching process is suppressed. By implementing additional physics, e.g. the erosion hindering effect of dilatancy and a proper bed slope effect, substantial improvements are achieved for idealised cases. However, two hurricane case studies showed that these model improvements do not hinder erosion sufficiently to achieve reasonable results. A proper description of bed roughness, which is preferably depth dependent and accounts for vegetation, together with calibration of the wave skewness and asymmetry is found to be very important. If this knowledge is applied on a newly introduced case study of Fire Island (hurricane Sandy, 2012), both breaching and overwash are modelled much more in line with reality. However, the complexity of having various morphodynamic processes within one model domain makes calibration a challenging task, requiring a more advanced bed roughness formulation.