In recent years, there has been an increase in storm surges along the coast of the UK that have resulted in white-water overtopping events. White-water overtopping events can generate large plumes of spray made up of a mixture of water droplets and air. These spray plumes can cause significant damage to the structures located landward of the coastal defences. Current design guidelines do not consider these types of events to be critical when designing structures located near the coastline. Moreover, there is very little research specifically focusing on downfall pressures caused by violent wave events. This thesis addresses this by means of experimental and numerical investigation into violent spray-dominated wave impact phenomena. An experimental campaign was carried out in The University of Manchester's wave flume for regular waves and focused wave groups impacting a model seawall located at the top of a sloping beach. Using pressure probes, the experiment measured pressures on the front face of the structure and on the deck of the structure. Standard filtering techniques were found to remove signal noise for pressure probes but also destroyed the peak pressures measured in the experiment. Using a wavelet filter was found to retain the measured peak pressures and remove signal noise enabling a wave-by-wave investigation into the spray-dominated impact events. For regular waves, significantly larger downfall pressures were measured on the deck of the structure than those previously reported in the published literature on the order of 20 to 40(rho)gHs. The peak downfall pressures were also measured further from the edge of the structure than previously reported at a distance of 2.33 Hs. The measured downfall events exhibit a range of behaviours from very short, high impact events to longer impacts of similar magnitude to the wave impact on the front face. A new correlation between wave period and the probability of the high deck pressures occurring has been proposed. Similarly, a new relationship between incident wave height and maximum deck pressures is proposed. Results for focused wave groups are less conclusive since the flume did not allow the generation of unbroken focused wave groups that generated spray. The highly nonlinear spray-dominated wave impacts on coastal structures are distorted and fragmented violent free-surface flows. To investigate the downfall impacts in more detail, a 2-D numerical model was developed with smoothed particle hydrodynamics (SPH) using the highly optimised weakly-compressible SPH (WCSPH) open-source code DualSPHysics accelerated on a graphics processing unit (GPU). Validations of the SPH model are presented comparing with experimental data published in the literature and from experiments carried out as part of this thesis. The validation studies produced results in general agreement with the experimental data for impact pressures on the vertical face and downfall impact pressures on the horizontal deck. The SPH model also reproduced mean uprush velocities with a 7% error. However, the single-phase SPH model could not reproduce the overtopping behaviour observed in the experiments. It is theorised that the lack of the air-phase in the model could explain why this occurred. However, using a multi-phase model with a uniform resolution is prohibitive in terms of computational expense due to the additional presence of the millions of air particles. A new multi-phase water-air variable resolution SPH model is therefore developed and implemented in the GPU DualSPHysics code. The new SPH model is validated using benchmark cases of still water and dry-bed dam break cases before being applied to a study of the Anchorsholme Seawall. Finally, application to a real-life test case is presented. The Anchorsholme seawall which failed during a storm in December 2013, was modelled using both the single phase and new multi-resolution, multi-phase SPH models. The wave condition used for the simulation was taken from the recorded sea-state that resulted in the shear failure of the recurve wall's bullnose. The results from the SPH model indicated that the wave loads were sufficient to have cause the failure of the bull nose. This is confirmed by subsequent structural analysis using the finite element (FE) software ABAQUS CAE. The FE model was first validated against existing experimental and field data on concrete and masonry structures subject to transient loads. The model adopted a dynamic explicit analysis and included material non-linearity.