Abstract: Under the background of climate change, the trend-caused nonstationarity of the hydrometeorological series in Huaihe River Basin are bound to impose a potential impact on the multivariable flood risk. Based on the LR-AIC criteria, this paper constructs a joint distribution model with dynamic Copula function as the core, takes the flood peak and flood volume series of Bengbu Hydrological Station in the Huaihe River Basin as the research object, and incorporates the urbanization factors and climate change factors related to the flood peak and flood volume series as covariates into the parameter structure of time-varying marginal distribution model and Copula model. The analytic results show that the proposed LR-AIC criteria for nonstationary detection can better capture the nonmonotonicity of extreme value series and the reverse trend of parameters than the nonparametric trend detection method. Due to the time-varying characteristics of maginal distribution parameters, the quantile pairs of peak flood volume at the level of 50-year joint return period show certain time-varying characteristics:（1） From 1960 to 1970, the design values of the peak flood volume showed an increasing trend；（2） From 1970 to 1990, it showed a downward trend；（3） From 1990 to 2010, it showed a gradually increasing trend. Only considering climate change or urbanization factors will lead to overestimation or underestimation of the design quantile of flood peak flood volume to a certain extent. Compared with urbanization factors, climate change factors are the dominant factors affecting flood univariate and multivariable risks in the sub-watershed controlled by Bengbu Hydrological Station. The nonstationary frequency analysis idea based on the dynamic Copula model is conducive to quantitatively isolating the impact of climate change factors and human activity factors（urbanization） on flood multivariate risk.