Does the spatial variability of the rainfall events affect the efficacy of the hydrological models?

The major socioeconomic issue affecting the nation's economy is flooding. The use of hydrologic models to predict floods has been the subject of numerous well-founded research, but little is still known about how spatial variability of rainfall affects the flood modeling. The purpose of this study is to comprehend, using the spatial variability index, how rainfall spatial variability affects flood characteristics and its importance in the selection of the hydrological models. The spatial variability index classifies the rainfall events into spatially homogenous (Class A) or spatially varying (Class B) by analyzing the spatial variability of rainfall and catchment properties. Further in this study, both lumped and distributed models have been set-up to understand whether the segregation of events prior to the flood modeling improves the model efficiency or not. Both the models were calibrated separately for the Class A and Class B events. The results shown that Class A events performed better in the lumped model with percentage error in peak flow (PEPF) of 18.45% and 20.32% in calibration and validation respectively. Whereas, the Class B events performance was better in the distributed model with PEPF of 11.5% and 15.85% in calibration and validation. These results were compared against the model calibrated using the traditional method, i.e. without segregating the events. The results show that performance of the lumped model is deteriorated. Similarly, distributed model performance was better when the class B events are separated instead of the combined events. Therefore, segregation of the rainfall events prior to the flood modeling helps in improving the model selection and its performance, which also reduces the assumptions in the model selection.