Variance-based Global Sensitivity Analysis of Surface Runoff Parameters for Hydrological Modeling of a Real Peri-urban Ungauged Basin

This paper proposes a new multi-step approach for sensitivity assessment of surface runoff parameters. The procedure has been tested on a peri-urban basin in southern Italy, interested by intense urbanization. The basin has limited data about land characteristics, and nearby precipitation measurements are not available. Accordingly, rainfall events are defined based on depth-duration-frequency curve valid for the area. The main novelties of the work are to provide a general framework for assessing the influence of runoff parameters (i.e. depression storage and surface roughness) for a basin model in SWMM in relation to rain events of various intensity/duration, and to provide a ranking of crucial parameters significantly affecting peak discharge and total volume of the hydrograph, for an ungauged basin, by means the Fourier Amplitude Sensitivity Test (FAST). Results indicate the dependence on rainfall characteristics of the relative importance of the parameters describing the pervious and impervious areas. Notably, the peak discharge of the shortest considered event is influenced only by the two parameters of the impervious area, while the opposite holds for the longest rain event. The total runoff volume is mostly influenced by the depression storage of impervious areas, with the parameters of pervious areas becoming more influential for longer rain events. Results allow a clear interpretation of the modelled physical processes variability within the basin and their relationship with rainfall/areas features, thus providing useful insights for key parameter definition in other contexts and for other models.