Derivation of flood frequency curves through a bivariate rainfall distribution based on copula functions: application to an urban catchment in northern Italy's climate

Abstract

The utilization of continuous approaches, namely analytical-probabilistic methods, has often been advocated for hydraulic device sizing, in order to overcome some deficiencies of the design event method. In the analytical distribution derivation, however, strong simplifying hypotheses are usually adopted. Rainfall depth and duration independency is the most unrealistic, even if it usually leads to satisfactory agreements between derived and benchmarking distributions. The reason can lie in drawbacks related to conventional assessment techniques of multivariate rainfall distributions. Copula functions recently provided a significant improvement in statistical inference capabilities and greatly simplified the distribution assessment. Nonetheless, the generalization of the return period concept, well defined in the univariate case, to multivariate cases has not found a blanket solution yet. Effective estimate methods can, however, be developed for the design and performance assessment of specific hydraulic devices. With regard to urban catchment applications, a criterion to derive flood frequency curves from a rainfall volume and duration distribution is herein proposed. Further, a method to estimate the return period of bivariate rainfall events based on a device-targeted approach is developed. Hydrologic simulations are conducted to support model reliability through a test case, featuring a northern Italian rainfall regime.