Coastal flooding events generated by the combination of different simultaneous Meteorological and Oceanographic processes that occur at different spatial and temporal scales, are a big concern for Pacific nations. Although for many applications, the full wave directional spectra aggregated into its bulk parameters (i.e., significant wave height, peak period and mean or peak direction) is used, this is especially unsuitable for islands located in large ocean basins where concurrent seas and swells approaching from every direction are common and including the directional complexity of the full spectrum becomes crucial. Moreover, when attempting to assess coastal flooding, it is imperative to analyze the setup and the energy released by infragravity waves at the coast in a high spatial and temporal resolution in order to generate the adequate conditions to force a flooding model. However, dynamically simulating the propagation of wave spectral energy to the coast in order to obtain setup and infragravity waves to feed a flooding model is a very computationally expensive task, particularly for Small Island Developing States (SIDS) where computational resources are often limited. For this reason, we propose the use of metamodels that combine statistical and numerical modeling to efficiently downscale waves and obtain the flooding extents needed to assess the impact associated to a particular event.