This paper presents a statistical analysis of storm surge in Hangzhou Bay, where storm surge is known taking place frequently. This study utilises measurement data taken the period of 1974–2006. The Pearson-III-Pareto distribution model was used first to fit with extreme values based on these measurement data and was then compared with Gumbel, Weibull, Pearson-III, Pareto distribution functions for estimations of extreme storm surge values correspondingly to 100-, 200-, 400-, 500-, 700- and 1000-year return periods. The predicted values of storm surge elevation in the 1000-year return period of the new model are 2.179% higher than the Weibull distribution and 3.546% lower than the Pearson-III distribution. Expectedly, the proposed Pearson-III-Pareto distribution is a more reasonable for presenting the statistical characteristic of extreme values of storm surge.

Extreme value; Extreme value theory; Measurement data; Pareto distributions; Return periods; Statistical characteristics; Storm surges; threshold

Ocean University of China; Xi′an University of Architecture ＆ Technology; Jiangsu University of Science and Technology; Zhejiang Ocean University