Wave growth and forecasting in variable, semi-enclosed domains

The accuracy of wave models in semi-enclosed-basins and orography-controlled wind conditions, especially during fetch-limited storm events, is known to be limited. Wind wave forecasting in the NW Mediterranean Sea is particularly demanding due to the characteristic sharp gradients of the wind and wave conditions. In this work we focus on the commonly observed underestimation of wave parameters even when the wind field is “correct” or overestimated. This is a small step to analyse such a discrepancy, where wind overestimation has been commonly used to get the “right” wave predictions for the “wrong” reason. Here we selected a suitable combination of nested meteorological and wave models to focus on the physics (in parameterized terms) of meso-scale wave generation in restricted domains. First, to better capture the typical sharp gradients in wind and wave fields under those conditions, the spatial resolution of the atmospheric model was progressively increased during a characteristic storm event from 18 km to 4 km; the corresponding frequency of the wind input was increased from 6 to 1 h. Second, the calculated rate of wave growth in the numerical model (i.e. the balance between the input term and the whitecapping dissipation) was analysed and tuned to match the observed local rate of wave growth. The rate of non-dimensional growth in the region of study, which was calculated using measurements along the fetch, turned out to be faster than simulated with the initial model settings and faster than reported in previous studies. Adjusting the wave growth rate in the model to the observations improved the estimated wave height by about 18% and the wave period by about 4%. Decreasing the grid size of the numerical models from 12 km to 4 km improved the timing of the wave peaks but not the maximum values of the storm. Increasing the frequency of the wind input (from 6 to 3 h) improved the estimation of the maximum wave height values (peaks) of the storm by about 13%. Summarizing, the results of this work showed that using high resolution and physically adjusted parameterizations in complex regions with strong wind and wave gradients such as the study area, it is possible to significantly reduce the under-estimation of wave parameters and to locally improve wave growth forecasting.