Assessing Meteo-HySEA Performance for Adriatic Meteotsunami Events.

Meteotsunamis are atmospherically-driven sea-level oscillations that can trigger hazardous coastal flooding, particularly in semi-enclosed and resonant harbors. Their accurate simulation and forecasting remain challenging because the ocean response depends critically on the intensity, propagation speed and spatio-temporal structure of mesoscale atmospheric pressure disturbances, which are often under-resolved even by state-of-the-art products.

This contribution evaluates Meteo-HySEA, a GPU-accelerated code designed for reproducing meteotsunami generation, propagation, coastal amplification and high-resolution inundation using a nested-grids approach. We benchmark Meteo-HySEA in the Adriatic Sea against the CPU-based AdriSC-ADCIRC modeling system for three well-documented events (June 2014, June–July 2017, May 2020), using WRF downscaling of ERA reanalyses and validation with high-frequency tide-gauge and microbarograph observations from the MESSI network complemented by additional coastal pressure records.

Results show that Meteo-HySEA generally reproduces the timing and spatial variability of simulated meteotsunami oscillations and often yields larger amplitudes than AdriSC-ADCIRC under identical forcing, while systematically overestimating dominant wave periods, especially in enclosed basins. For the 2017 and 2020 events, both modeling frameworks significantly underestimate observed amplitudes at key hotspots (e.g., Vela Luka and Stari Grad), consistent with deficiencies in the modeled atmospheric disturbances, highlighting atmospheric forcing as the dominant source of uncertainty. Controlled synthetic-pressure experiments further indicate systematic differences in energy trapping and damping within harbors, emphasizing sensitivity to nearshore resolution, dissipation/parameterizations, the treatment of wetting–drying fronts and inundation.

Crucially, GPU acceleration enables order-of-magnitude gains in computational efficiency, supporting rapid high-resolution simulations and making Meteo-HySEA a strong candidate for ensemble-based meteotsunami forecast, extending the modeling chain from offshore oscillations to onshore flooding. This functionality is particularly relevant for risk assessment and civil protection, as it allows the estimation of direct impacts on vulnerable harbors and urban waterfronts.