For accurate simulations in SWMM, model parameters need to be calibrated. A double calibration was performed in the presented research. It involved both the RUNOFF module and the GRs LID module, by a comparison of modeled and observed streamflow (RUNOFF module) and runoff (LID module). The optimal parameters were selected corresponding to the maximization of the NSE (Nash–Sutcliffe Efficiency Index) (Nash & Sutcliffe 1971) given by the following equation:
where N represents the length of the sample, i is the time index in an hour, SIM and OBS, respectively, represent the predicted and measured values, and is the mean of the measured values. The calibration of the RUNOFF module was performed by comparing modeled and observed streamflow at the event hourly scale, at the outlet of the Sarno Mouth sub-catchment (Figure 1(a)). The selection of the calibration parameters was made in accordance with Behrouz et al. (2020). Based on the review of several previous studies, they suggested that parameters exhibiting the largest sensitivity, or that were most frequently reported as sensitive in the literature, were the catchment imperviousness, the width, the impervious depression storage coefficient and the channel Manning's roughness coefficient. In the present research, the percentage of imperviousness is a known parameter deriving from SAR image elaboration while the width has been obtained by dividing the catchment area by the average maximum overland flow length as suggested by the SWMM user's manual. Consequently, the impervious depression storage coefficient dp and channel Manning's roughness coefficient n were considered as the only parameters to be calibrated. These parameters are supposed to vary uniformly across the basin and so they assume the same value for each sub-catchment and model link. The range of expected values for each parameter, which considers the physically plausible values, has been shown in Table 5, according to the review of previous studies (Barco et al. 2008; Rossman 2010; Sadeghi et al. 2022).