Sensitivity Analysis and Calibration of a Rainfall-Runoff Model with the Combined Use of EPA-SWMM and Genetic Algorithm

• Published in: Acta geophysica Vol. 64; no. 5; pp. 1755 - 1778
• Main Authors: Del Giudice, Giuseppe, Padulano, Roberta
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.10.2016; De Gruyter; Springer Nature B.V
• Subjects: Algorithms; Calibration; Concentration time; Earth and Environmental Science; Earth Sciences; genetic algorithm; Genetic algorithms; Geophysics/Geodesy; Geotechnical Engineering & Applied Earth Sciences; identification analysis; Rainfall distribution; Rainfall measurement; rainfall-runoff model; Rainfall-runoff relationships; Runoff; Runoff coefficient; Sensitivity analysis; Sewers; Storage coefficient; Structural Geology; genetic algorithm; sensitivity analysis; calibration; identification analysis; rainfall-runoff model
• ISSN: 1895-6572; 1895-7455; 1895-7455
• DOI: 10.1515/acgeo-2016-0062

An integrated Visual Basic Application interface is described that allows for sensitivity analysis, calibration and routing of hydraulichydrological models. The routine consists in the combination of three freeware tools performing hydrological modelling, hydraulic modelling and calibration. With such an approach, calibration is made possible even if information about sewers geometrical features is incomplete. Model parameters involve storage coefficient, time of concentration, runoff coefficient, initial abstraction and Manning coefficient; literature formulas are considered and manipulated to obtain novel expressions and variation ranges. A sensitivity analysis with a local method is performed to obtain information about collinearity among parameters and a ranking of influence. The least important parameters are given a fixed value, and for the remaining ones calibration is performed by means of a genetic algorithm implemented in GANetXL. Single-event calibration is performed with a selection of six rainfall events, which are chosen so to avoid non-uniform rainfall distribution; results are then successfully validated with a sequence of four events.