An efficient causative event-based approach for deriving the annual flood frequency distribution

•An efficient and robust method for the estimation of the annual flood frequency distribution is developed.•It combines the advantages of the continuous simulation and event-based methods, i.e., accuracy and efficiency.•The joint probability nature of the rainfall-runoff process is taken into account and the AEP neutrality assumption is avoided.•It approach uses the causative events of the streamflows to derive the annual flood distribution.•It reduces computational demand compared with a long continuous simulation.

SummaryIn ungauged catchments or catchments without sufficient streamflow data, derived flood frequency methods are often applied to provide the basis for flood risk assessment. The most commonly used event-based methods, such as design storm and joint probability approaches are able to give fast estimation, but can also lead to prediction bias and uncertainties due to the limitations of inherent assumptions and difficulties in obtaining input information (rainfall and catchment wetness) related to events that cause extreme floods. An alternative method is a long continuous simulation which produces more accurate predictions, but at the cost of massive computational time. In this study a hybrid method was developed to make the best use of both event-based and continuous approaches. The method uses a short continuous simulation to provide inputs for a rainfall-runoff model running in an event-based fashion. The total probability theorem is then combined with the peak over threshold method to estimate annual flood distribution. A synthetic case study demonstrates the efficacy of this procedure compared with existing methods of estimating annual flood distribution. The main advantage of the hybrid method is that it provides estimates of the flood frequency distribution with an accuracy similar to the continuous simulation approach, but with dramatically reduced computation time. This paper presents the method at the proof-of-concept stage of development and future work is required to extend the method to more realistic catchments.