Optimization of sizing of annual water storage reservoirs considering return period association.

Abstract

An increasing concern among water resources managers is the search for ways to improve reservoir sizing techniques. A simple technical and scientifically based technique is the sequent peak method, which has been widely disseminated, but presents limitations with respect to the length of the available data series. The objective of this study was to propose modifications to the sequent peak method to overcome the limitations related to the impossibility of associating the reservoir storage capacity with return periods different from the length of the data series. Storage capacities were estimated for every year, considering gauge stations located in the Urucuia River Basin, based on the sequent peak method. In order to associate such capacities with a frequency factor (return period), the Gumbel distribution was applied to the estimated storage values. This association with return periods which differ from the number of years in the data series showed great applicability.