The Influence of the Annual Number of Storms on the Derivation of the Flood Frequency Curve through Event-Based Simulation

This study addresses the question of how to select the minimum set of storms that should be simulated each year in order to estimate an accurate flood frequency curve for return periods ranging between 1 and 1000 years. The Manzanares basin (Spain) was used as a study case. A continuous 100,000-year hourly rainfall series was generated using the stochastic spatial–temporal model RanSimV3. Individual storms were extracted from the series by applying the exponential method. For each year, the extracted storms were transformed into hydrographs by applying an hourly time-step semi-distributed event-based rainfall–runoff model, and the maximum peak flow per year was determined to generate the reference flood frequency curve. Then, different flood frequency curves were obtained considering the N storms with maximum rainfall depth per year, with 1 ≤ N ≤ total number of storms. Main results show that: (a) the degree of alignment between the calculated flood frequency curves and the reference flood frequency curve depends on the return period considered, increasing the accuracy for higher return periods; (b) for the analyzed case studies, the flood frequency curve for medium and high return period (50 ≤ return period ≤ 1000 years) can be estimated with a difference lower than 3% (compared to the reference flood frequency curve) by considering the three storms with the maximum total rainfall depth each year; (c) when considering only the greatest storm of the year, for return periods higher than 10 years, the difference for the estimation of the flood frequency curve is lower than 10%; and (d) when considering the three greatest storms each year, for return periods higher than 100 years, the probability of achieving simultaneously a hydrograph with the annual maximum peak flow and the maximum volume is 94%.