Estimating Extreme Water Levels With Joint Probability Optimal Sampling Method In Xiamen Island

Xiamen Island is located on the southeast coast of Fujian province,and is close to the Taiwan Strait.Strong typhoons are frequently happened near Xiamen Island,which caused storm tides and threatened the security of Xiamen Island.To predict the extreme water level caused by storm surge is of great significance to the flood control and disaster prevention and offshore engineering construction of Xiamen Island.In this paper,a method that combines JPM-OS model and ADCIRC-SWAN hydrodynamic model were established,which were used to predict the extreme water level of Xiamen Island under the influence of storm surge.The main contents of this study are hydrodynamic model establishment and verification,joint probability optimal sample method(JPM-OS)and typhoon recombination,storm surge simulation of recombinant typhoon,and extreme water level prediction.The main work and conclusions are as follows:(1)ADCIRC model was set up for research area and astronomical tide simulation verification was implemented.The Fujita-Gaoqiao formula combined with the gradient wind pressure field and Mr.Ueno Takeo formula were used as the typhoon wind model.The wind fields of the No.1011 Typhoon "Fanapi" were simulated,the wind speed and wind direction on the site with actual observation data were verified.ADCIRC-SWAN model was used to simulate storm surge water level process caused by No.1011 Typhoon.The water levels of Pingtan,Chongwu,Xiamen and Dongshan station were compared with the actual observation data,the verification results were reasonable.(2)Based on historical typhoon data,analyzed typhoon samples,with the limited parameters combination as a typhoon and allocated an appropriate weight for each typhoon.Optimum sampling of typhoon samples was performed by Bayesian orthogonal algorithm.The multidimensional integral was converted into a weighted sum to reduce the computational burden.The JPM-OS method was implemented to reconstruct the typhoon.First,71 typhoon samples were collected.Four main characteristic values(central pressure,landing speed,Angle and the typhoon landing site to coastline reference center distance)in the typhoon landing site were treated as parameters for probability distribution fitting analysis.Then,the optimal combination sampling and joint probability calculation of the main characteristics of the typhoon were carried out using the Bayesian orthogonal algorithm.Each characteristic parameters related to the distance parameter were assigned as weight.70 typhoons were restructured by the optimal sampling method with each probability of occurrence and matched three typical paths.(3)The synthetic typhoons obtained by JPM-OS method were used as the driving factor of storm surge,based on astronomical tidal level,the water dynamic calculations were carried out by ADCIRC-SWAN couple model.The tidal data of four stations were extracted,and the cumulative probability and extreme water levels of four stations were collected.The prediction for 50-years,100-year and 200-year annual maximum water levels in Gulangyu station were compared with the traditional frequency analysis forecast results,the relative error less than 1.83%,which means the verification results were good.A whole set of JPM-OS method were feasible in Xiamen extreme water level prediction research,thus the multiple sites around xiamen island about 100-year and 200-year annual maximum water levels were predicted,in which the 100-year and 200-year annual maximum water levels in Gulangyu station were 7.78 m ? 7.88 m,respectively.