flood frequency analysis using density function of wavelet (case study: polroud river)

in this paper we present a method to perform flood frequency analysis (ffa) when the assumption of stationary is not important (or not valid). a wavelet transform model is developed to ffa. a full series is applied to ffa using two different wavelet functions, and then a combined method is investigated. in the combined method, all discharge data which were less than the lowest value of annual maximum (am) discharge were removed. furthermore, energy function of wavelet was used for ffa. the data was decomposed into some details and an approximation through different wavelet functions and decomposition levels. the approximation series was employed to ffa. this was performed using discharge data from of the polroud river in iran. this paper analysis was performed on the daily maximum discharge data from the tollat station in the north of iran. data from 1975 to 2007 was evaluated by wavelet analysis. the study shows that wavelet full series model results (density function) are too small in compared with the results of combined method and they are both lesser than traditional methods (am and pd). in other hand the results of energy function method is closed to the combined method when they are compared with the full series data results. these wavelet models were assessed with the am and pd methods. the concrete result of this paper is that, the basin hydrologic conditions and data's nature are very important parameters to improve ffa and to select the best method of analysis.