Probability Analysis on the Typhoon Induced Sea States of the South China Sea

The gas and oil exploitation in the South China Sea has gradually accelerated in recent years. Some platforms, such as Liwan3-1, Panyu30-1, etc, are put into use. And there are many deep-sea platforms being built. As known, the marine engineering structures are impacted by complicated sea environment. Especially in typhoon season, the extreme sea states induced by typhoon may lead to structural failures. So probability analysis of the typhoon induced sea states is important to marine engineering designing. The typical method is using some probability distribution models to fit the observed or hind-cast data for calculation of combination of the sea states with different return periods. In this process Gumbel distribution, Fréchet distribution, Weibull distribution, Log-normal distribution, P-III distribution are commonly used. There are some problems that need to be solved. Firstly, all of the above models are based on yearly maximum sampling method, so the length and quality of data series will influence the probability analysis obviously. Sometimes the observed data of typhoon induced sea states is not enough for getting the results of high quantiles. Secondly, using different probability models, all of which can pass the hypothetical testing, will get different analysis results. The method differences of parameter estimation will also lead to deviation of probability prediction. In this paper a Double-Layer Nested Probability Model (DLNPM) is proposed based on the Multivariate Compound Extreme Value Distribution (MCEVD) and applied to the long term probability analysis of typhoon induced extreme sea environments of the South China Sea. And the uncertainty analysis caused by different methods and data series is carried out in the study. In the new model typhoon characteristics, including maximum central pressure difference ( ΔP), radius of maximum wind speed (Rmax), moving speed of typhoon center ( Vf), minimum distance between typhoon center and target site ( δ), typhoon moving angle ( θ) and annual typhoon frequency ( λ), are taken into account. The hind-cast data are calculated based on the joint probability analysis of these typhoon characteristic factors. And the long term probability prediction considering uncertainties are gotten.