Joint risk analysis of typhoon hazards based on coupled ADCIRC-SWAN model simulations around Hainan, China

With rising sea level and increasing typhoon intensity under greenhouse warming, risk of typhoon-related hazards, including strong winds, storm surges and waves in coastal regions, is becoming increasingly severe. Hainan province of China, adjacent to the western North Pacific, is vulnerable to typhoon hazards. In this study, we constructed 65 typhoons using the Jelesnianski model to force the Advanced Circulation (ADCIRC)-Simulating Waves Nearshore (SWAN) coupled model and reproduced typhoon waves and surges, then assessed their extreme values around Hainan under 20-, 50-, and 100-year return periods. Taking Sanya station as an example, we further conducted joint risk analysis via considering bivariate and Trivariate hazard factors. The relative errors of modelled wind speed (V), significant wave height (SWH), and storm surge (S) are less than 10 % compared to in-situ observations. Using the Gumbel function, extreme values of V, SWH, and S under the 100-year return period around Hainan province can exceed 40 m s −1 , 15 m, and 2 m. After Kolmogorov test, the Gumbel-Hougaard Copula function was used to estimate bivariate and trivariate joint probability distributions at Sanya station and calculate the “OR” return periods of different joint events. The “OR” return period of three factors at the Sanya station under the 100-year extreme value is 48.79 years, indicating a high probability of simultaneous occurrence on account of their correlation. This study focused on the complex topography surrounding Hainan Island and utilized a coupled model to compensate for the limitation of synchronous observational data in typhoon joint risk analysis. • The ADCIRC-SWAN model was used to simulate surges and waves under 65 typhoons influencing Hainan province from 1990–2021. • For a 100-year return period, wind speed, significant wave height, and storm surge exceed 40 m s -1 , 15 m, and 2 m in Hainan. • Joint risk analysis indicates trivariate factors show a higher probability of occurrence than single or bivariate factors.