The return values of storm surge elevation are frequently estimated by the annual maximum value method assuming all individual data vests in a common parent distribution disregarding the fact that the inducements of the storm surge are considerably changeable. This paper examines the magnitude of the error in estimation of return values owing to this assumption. The storm surge data measured at Dagang Tide Station of Qingdao, China within the last thirty years is used to set up different statistical samples according to multiple time divisions, such as year, season or month. A new statistical model considering seasonal variation of the storm surge is put forward to compute the return value. The calculation results obtained by annual maximum value method and peak over threshold (POT) method are compared and discussed. The existing error in the calculation of return value, compared with the proposed model, indicates that the assumption ignoring seasonal variation leads to underestimate of return values. Return values of storm surge elevation in the coastal area of Qingdao are provided for engineering design using the new statistical model.
In recent years, the natural hazards triggered by storm surge have been increasingly impacting on the coastal areas of China. Storm surge is the uncommon piling up of the seawater along the shoreline induced by intense atmospheric interference such as tropical or temperate cyclones. Zhao et al. (2000) presented the long-term forecast of the return value for storm surges in Chengbei area of the Bohai Sea in China, where as a matter of fact two dominant weather conditions might cause the storm surge, the northward typhoon and the cold waves generated by storms caused by passing cold fronts during the winter, early spring and late fall, and as well accompanied with strong winds (Zhou, 2000).
