Process-based numerical models are essential tools for predicting nearshore hydrodynamics and supporting climate-resilient coastal management. However, the predictive capability of these models strongly depends on site-specific calibration, while robust and transferable calibration methodologies remain limited. This study develops and validates a systematic calibration framework for the XBeach model in 2DH surfbeat mode for a sandy coastal environment in the northern Adriatic Sea. Offshore wave conditions measured by a directional wave buoy were used to force the model, whereas calibration relied on concurrent hydrodynamic observations collected by two Acoustic Doppler Current Profilers (ADCPs) deployed in shallow water. Three storm events with contrasting meteorological forcing, including two Bora storms and one Scirocco storm, were selected to ensure calibration under a broad range of wave conditions. A structured sensitivity analysis was performed by progressively adjusting the parameters controlling wave transformation, bottom friction, energy dissipation and directional wave propagation. Model skill was quantified through a normalized multi-metric performance framework integrating RMSE, bias, MAE and Pearson's correlation coefficient for significant wave height, current velocity and current direction, allowing an objective comparison among calibration scenarios. Results demonstrate that model performance is governed by a limited number of highly influential parameters. In particular, the wave friction coefficient (fw) proved to be the dominant control on wave transformation, with an optimal value of 0.02 substantially improving wave-height predictions. Explicit directional wave propagation (single_dir = 0) significantly enhanced current-direction simulations, whereas activating wave–current interaction (wci = 1) consistently reduced model skill. The calibrated configuration achieved an overall normalized score of 0.216 and successfully reproduced two independent storm events, yielding validation scores of 0.365 and 0.183, thus demonstrating the robustness and transferability of the proposed parameter set across different hydrodynamic conditions. Beyond the specific study site, the results highlight the limitations of adopting calibration parameters directly from the literature and show that reliable process-based modelling requires a site-specific yet systematic calibration strategy. The proposed framework provides a reproducible methodology for calibrating XBeach in sandy coastal environments and offers new insights into the relative importance of wave dissipation and directional processes in improving nearshore hydrodynamic predictions.