By coupling the Critic weight method, local comparison method and fuzzy set analysis method, an improved fuzzy set analysis method for reservoir flood season staging is proposed. This method converts multiple indicators into a comprehensive index using the Critic weight method, and then use the local comparison method to analyze the fluctuation degree of the comprehensive index within the step size to objectively determine the flood season staging nodes. Subsequently, the staging results of the improved fuzzy set analysis method are compared with the results of other staging methods to determine the optimal flood season staging scheme. On this basis, the P-Ⅲ distribution, two-parameter Gamma distribution, log-normal distribution and generalized extreme value distribution are used to fit the marginal distribution of design elements and determine the optimal distribution line. The joint distributions of flood peaks and flood volumes in different periods are constructed by the Frank Copula function, the design frequency for each period is deduced, and the design values of the design elements and the design flood hydrograph for each period are determined. The proposed method is applied to the flood season staging and staged design flood calculation for the Longyangxia Reservoir. The flood season staging scheme of the Longyangxia Reservoir is determined as follows:the early flood season is from early June to late June, the main flood season is from early July to late August, and the late flood season is from early September to late October. The calculated design flood hydrograph of each stage at the Longyangxia Reservoir shows that the main flood season is characterized by mostly sharp and thin flood hydrographs, and the early and late flood seasons are characterized by mostly short and fat hydrographs. The flood peak and volume in the main flood season are greater than those in the early and late flood seasons. The rationality test of flood season staging shows that the flood season staging nodes determined by the improved fuzzy set analysis method are more reasonable than those determined by the traditional methods.