How many storms are needed to estimate the maximum annual flood? A continuous, fully-distributed, physically-based model approach

In order to estimate extreme flood hydrographs accurately by stochastic approaches, it is necessary to carry on hydrologic simulations of thousands of years. By using continuous fully distributed simulation models we are able to derive flood frequency distributions from the continuous hourly streamflow obtained at any desired point in the drainage network of the basin. This approach has the advantage of estimating the variables for the entire period of simulation. However, continuous models tend to be more complex than event-based models, with computational efforts that could be very intensive, even by using high performance computing and parallelization processes. Event-based simulations require much shorter simulation times. Among other disadvantages, the maximum annual rainfall does not necessarily have to produce the maximum annual flood. Furthermore, different properties of storm events could affect in the derivation of flood frequency curves: rainfall temporal distribution, event duration, maximum intensity, total storm depth, etc. Therefore, the estimation of high return period flood frequency curves by using event-based models is a challenge. This paper focuses on addressing this challenge by analysing the results obtained by coupling a continuous stochastic weather generator with a continuous distributed physically-based hydrological model. The objective is to propose a criterion for the selection of the minimum number of storm events per year to guarantee a correct derivation of the flood frequency curve for high return periods, according to its maximum annual peak-flow, maximum annual volume, and the bivariate analysis of both variables. The outcomes of this research would be useful for event-based model applications.