The Beibu Gulf is located in the southwest of China, to the northwest of the South China Sea. This gulf is a typical semi-closed sea area and lies on the continental shelf. The mean water depth within the Beibu Gulf is about 46 m and the maximum value is no more than 100 m (Chen et al., 2011; Zhou et al., 2015). This gulf is also perennially exposed to the northeast and southwest monsoonal influences and typhoons pass by almost every year. The Beibu Gulf is managed jointly by China and Vietnam and is one of the best fishing grounds for the former and an important seaport in southwest China (Shao et al., 2018; Yu & Mu, 2006). Ocean waves exert significant impacts on fishing and oil platform safety within the Beibu Gulf (Yan et al., 2013, 2015). Waves are also essential factors for inducing disasters, based on wave data obtained from buoys within the Beibu Gulf between 2015 and 2018 and data show that ocean wave disasters refer to the wave height up to 6 m and mainly induced by typhoons and cold air took place 2.5 times per year on average and even four times in 2016 in the Beibu Gulf (Su et al., 2019). Wave height is an essential element that can be used to describe waves; its monitoring and acquisition are of great significance. Few studies have so far been performed on wave height prediction for the Beibu Gulf area. Studies aimed at forecasting significant wave height (SWH) within the Beibu Gulf are vital for fishery breeding and boat travel (Chen et al., 2009; Fu et al., 2019). Ocean waves are common phenomena and vital components of physical oceanography. External factors, including kinetic and physical processes and boundary conditions, all affect the generation, development, and extinction of ocean waves. Several numerical models have, therefore, been established and used to calculate and predict waves state. The WAM model, a third-generation wave approach, was proposed in 1988 for ocean wave prediction (Group, 1988). Subsequent to the development of computer technology, WAVE WATCH III was then presented based on the WAM model (Tolman, 2009) and was applicable to open ocean field predictions. Another classical third-generation wave model, simulation waves near shore (SWAN), was also applicable to wave prediction in coastal regions (Booij et al., 1999). These three models all exhibit excellent performance in wave height forecasting (