Improvement of wave predictions in marginal seas around Korea through correction of simulated sea winds

One of the main error sources in predicting ocean surface waves is the uncertainty of numerical sea wind products used as input to wave prediction models. In particular, the complexity of atmospheric physics and limitations of numerical computation in time and space increase the uncertainty of numerical sea winds in the area of marginal seas. So, a process is proposed to improve wave prediction in marginal seas by correcting numerical sea-wind products (i.e., herein, ERA5 wind reanalysis data produced around the Korean Peninsula). To evaluate and correct the ERA5 data, Korea Meteorological Administration (KMA) buoy and MetOp-B (ASCAT) satellite data were used. In the procedure, firstly, the ASCAT winds in space were calibrated using the buoy data, and secondly the calibrated satellite winds were used to evaluate and correct the ERA5 wind data. Finally, the corrected ERA5 wind data were used as input to a numerical wave model (i.e., WAVEWATCH-III). When the corrected wind data were applied to the wave numerical model, the resulted predictions showed a better match to the in situ wave measurements. In the results, the bias error was reduced from (0.21 to 0.04) m, and the best fit slope was improved from (0.84 to 0.93) on average, compared to the predictions by the original sea winds. This procedure can be applied to other marginal seas in which sea winds are strongly influenced by neighbouring continents, and not well reproduced by numerical approaches.