Simulation of hydrograph is important in planning measures against sediment and floods damages caused by heavy rainfall. However, since the applicability of the parameters to disasters of different scales is not clear, the issue is whether large‐scale disasters can be simulated with the parameters identified from the past data of small and medium‐sized floods. Taking the sediment and flood damage in Marumori Town, Miyagi Prefecture in 2019 as an example, we identified the parameters p and k of the storage function method for small and medium‐scale floods and examined the reproducibility for large‐scale floods. Recent hydrological studies suggested that if the rainfall magnitude became large, the watershed should be the wettest condition and the storage‐runoff function became close to catchment‐specific relationship. Thus, here we hypothesized that if we used relatively large rainfall data to calibrate parameters, the calibrated parameters might be effective for prediction of hydrograph in the extremely large rainfall event, although the calibrated rainfall event on a scale was not large enough to cause a disaster. When the calibrated p was 0.2 to 0.3, the fluctuation of the observed hydrograph and the peak discharge could be reproduced successfully. However, some of calibrated parameter sets did not work well. This suggested that the wetness of catchment was not sufficient for the entire catchment area to contribute to the runoff in these rainfall events. Therefore, we conclude that calibrating the parameters with relatively large events may predict hydrographs with extremely large events, but to obtain reasonable parameter values, we still need to pay attention to the wetness contribution area.