Determination of water level design for an estuarine city

Based on the extreme value theory, self–affinity, and scale invariance, we studied the temporal and spatial relationship and the variation of water level and established a model of Gumbel–Pareto distribution for designed flood calculation. The model includes the previous extreme value models, the over–threshold data, and the fractal features shared by previous extreme value models. The model was simplified into a logarithmic normal distribution and a Pareto distribution for specific parameter values, and was used to calculate the designed flood values for the Shanghai Wusong Station in 100– and 1 000–year return periods. The calculated results show that the value of the designed flood height calculated in the Gumbel–Pareto distribution is between those in the Gumbel and Pearson–III distributions. The designed flood values in the 100–and 1 000–year return periods of the model were 0.03% and 0.11% lower, respectively, than the Gumbel distribution and 0.06% and 1.54% higher, respectively, than the Pearson–III distribution. Compared to the traditional model based solely on extreme probability, the Gumbel–Pareto distribution model could better describe the probabilistic characteristics of extreme marine elements and better use the data.