This article comparatively analyzed the four designed flood recurrence levels of bivariate flood joint distribution of flood peak discharges and flood volumes by using Archimedean copulas function

Kendall distribution function and survival Kendall distribution function. The annual maximum flood peak discharges and the corresponding flood volumes of Qilinzui hydrometric station at Zengjiang River as an example

the ‘OR’ primary return periods

‘AND’ primary return periods

Kendall return periods and survival Kendall return periods of bivariate flood joint distribution and the most likely design flood quantiles were computed by using the optimally fitted Gumbel-Hougaard copula. The main conclusions of this study can be summarized as follows: comparing the different specific flood return periods

the Kendall return periods can be more accurately reflect the flood risks relative to ‘OR’ primary return periods; while the survival Kendall return period can be more accurately reflect the flood risks when two variables for flood were supercritical at the same time with respect to ‘AND’ primary return periods. The univariate flood design value has reached safety standards according to the relevant specifications of the current design requirements. The estimated flood design quantiles of the ‘OR’ primary return periods and two variables with the same frequency were obviously higher than the single variable floods. The mostlikely design realizations of the Kendall return periods and the survival Kendall return periods can serve as a new selection for flood control projects safety and risk management.