Enhancement of flood warning system performance using stochastic rainfall threshold curve

Flood warning systems are the important and effective approaches to prevention or mitigation life and property loss in any flood event. All flood warning systems use one or more flood event indicator such as climatological signals. Rainfall threshold curve is one the conventional flood events indicator in many flood warning systems. In this respect, observed or forecasted rainfall was compared with rainfall threshold to notify flood event. Application of some assumptions like constant rainfall patterns and rainfall-runoff model parameters such as loss and base flow is known as the main drawback of using conventional rainfall threshold curves. This study has considered uncertainties of rainfall-runoff model parameters and variables for extracting rainfall threshold curve, while solving shortcomings of the previous works. Results of this study demonstrated that extraction of the rainfall threshold curves by considering uncertainties of the mentioned variables and parameters represents a very higher accuracy respect to conventional rainfall threshold curves.