2020 年 76 巻 2 号 p. I_67-I_72
A two-dimensional XBeach model faces difficulty in updating the bathymetry profile under calm conditions. This study investigates the inaccuracy of predicting the undertow, which is a seaward-directed current affecting the suspended sediment load. Progress was made in applying XBeach to the prediction of the observed undertow in the field in the Hasaki coast of Japan. The observations were performed between May 9 and June 2, 2016. To assess the capability of XBeach to reproduce the undertow under low-wave-energy conditions, the datasets observed during the calm situation (after May 24, 2016) are used, and the parameters sensitive to the undertow prediction are adjusted. The results reveal a time-shift problem that occurs at the time-varying undertow velocity. The comparison becomes better when the undertow is forward-shifted for 3 h but remains underestimated. To improve the XBeach default results, a modification is implemented based on the Stokes drift, using the coefficient of water depth. This approach improves the accuracy of undertow prediction in XBeach from bad to reasonable quality for the average of RMAE value (from 0.77 to 0.48), against the observation.