Performance evaluation of SWAN ST6 physics forced by ERA5 wind fields for wave prediction in an enclosed basin

Abstract

In this study, the performance of SWAN (Simulating WAves Nearshore) spectral wave model ST6 physics forced by ERA5 reanalysis wind data were evaluated in predicting the wave properties in an enclosed basin, the Black Sea. ERA5 wind data were validated against in-situ measurements. Although two data match quite well, a 10-20% underestimation in wind speeds and velocity components was found. Spectral wave modeling study was conducted in two phases. In the first phase, 16 model setup combinations were generated using default parameterizations of SWAN spectral wave model with different physics (Westhuysen, Komen, Janssen, ST6) and with different wind sources (ERA5, ERA-Interim, and CFSv2). The model setups forced by ERA5 winds had the highest correlation values and in combination with ST6 physics performed better than the others in the prediction of wave heights. At the next stage, this model setup was calibrated at two stations by conducting 82 different parameterizations. Chosen model parameterization with overall satisfying performance was validated at five buoys. Comparisons among similar runtime model runs with timesteps up to 60 min and iteration steps up to four, show that higher timesteps with higher iterations give better results than lower time steps with one iteration. Debias and wind scaling parameters were found to be the most significant at adjusting wave heights. Comparison with previous calibration studies indicated that model runs based on calibrated ST6 physics parameterization and forced by ERA5 winds provided an increased accuracy of wave height predictions in the Black Sea.