Estimation of extreme wind speeds with different return periods in the Northwest Pacific

It is vital to analyze extreme wind speed in marine engineering designs. However, due to the lack of observational data, it is impossible to establish the measured long-term wind speed series. This study simulates the annual hourly wind field of every tropical cyclone (TC) with a resolution of 5 km in the Northwest Pacific (NWP) from 1981 to 2020. On this basis, combined with the sea surface wind speed data observed by the satellites and the ships, the 40-year annual maximum wind speed series of NWP are established. The Gumbel, three-parameter Weibull (Weibull-3par), two-parameter Weibull (Weibull-2par), generalized extreme-value (GEV) distribution, and the two parameter estimation methods are used to estimate the extreme wind speeds with different return periods (RPs) at four typical locations in the NWP. Meanwhile, the effects of different extreme-value distributions and different parameter estimation methods on the estimation results are discussed. Subsequently, the best distribution and parameter estimation method for each grid in the NWP are determined by the goodness-of-fit test, and then the spatial distributions of extreme wind speeds with different RPs along with uncertainty estimates in the entire NWP are obtained. The results show that extreme wind speeds with RPs of 5, 25, 50, and 100 years in the east of Taiwan and Philippines can reach a maximum of 43.8, 60.8, 70.4, and 81.4 m s⁻¹, respectively.