Abstract: In the aim of providing basis for the operation of sandy reservoir and reducing the influences of length and randomness of flow and sediment series, wavelet transform was adopted to analyze the variation characteristics of flow and sediment series. Standardized monthly flow and sediment series were decomposed by Multi-resolution Analysis using the db3 wavelet function, and continuous wavelet transform was used to evaluate the periodic variations of the two standardized series using the complex valued Morlet function. The flow and sediment series at the dam site of Yayangshan Hydropower Station located in Lixianjiang watershed were taken as an example. The reconstruction of the lowest frequency part revealed the trend of the flow and sediment series. According to the local rainfall and soil and water conservation, the result is considered reasonable. Wavelet variances were obtained to identify the dominant period as 2-year approximately for both flow series and sediment series and reveal the synchronization between them. The results indicate that wavelet transform is effective for nonstationary stochastic series analysis.

Key words: wavelet transform, flow and sediment series, runoff, sediment concentration, time scale, multi-resolution analysis, wavelet variances