Role of underlying surface, rainstorm and antecedent wetness condition on flood responses in small and medium sized watersheds in the Yangtze River Delta region, China

Floods are becoming more frequent and hard to control due the shortage of water conservancy projects in small medium sized basins, especially developing regions. Understanding hydrologic responses estimated flood characteristics storm events can help predict disasters form better effective mitigation adaptation strategies. Using observations several representative watersheds Artificial Neural Networks (ANN) Principal Component Regression (PCR) models, we conducted data-driven analyses examine responding characteristics. We quantified relative contributions influencing factors variation each characteristic. Statistical analysis shows that drainage area plays a key role determining distribution lag time peak discharge. Rainstorm variability has direct influence on floods, typhoon-induced rainstorms with high total rainfall intensity generate higher time, peak, unit discharge runoff depth, but lower coefficient. The ANN PCR models accurately predicted variations features using driving including physical geographical characteristics, rainstorm features, antecedent condition. Physical geographic influential coefficient, while magnitude depth. These results indicate floods mainly affected by Yangtze River delta, it might become damaging increasing extremes sprawling impervious surfaces changing environment.