Impact of the Initial Conditions from an Atmospheric Model on a Wave Forecast System

Abstract

This study aimed to improve the accuracy of dangerous wave prediction, using the example of the Korean east coast. Generally, the accuracy of wave prediction depends on the accuracy of weather prediction. Therefore, we attempted to improve the performance of a weather model by improving its initial conditions. In this study, two initial weather conditions obtained from an atmospheric weather model were employed, to simulate high, swell-like waves, in the East Sea. In the first simulation, the Weather Research and Forecasting (WRF) model was run for 72 h, from a cold-start, cycling a 3D-Var data assimilation through, every 3 h, from the start time to +6 h (a "cold-start" run). In the second simulation, the WRF model was run for 78 h from cold-start, cycling the 3D-Var data assimilation every 3 h from -6 h to +6 h, in a simulation called a "hot-start" run. The WRF model was initialized using boundary conditions from the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS), and SWAN 41.10 was used for the wave model. Based on observational data collected over six months, the error in the "hot-start" prediction run was reduced by approximately 5%, in comparison with the "cold start" run.