Evaluation of Trends and Dominant Modes in Maximum Flows in Turkey Using Discrete and Additive Wavelet Transforms

Abstract

This paper aims to define trends and dominant modes in annual instantaneous maximum flows (AIMF) covering the period 1961-2014 from 10 gauge stations located in different river basins in Turkey. To achieve this aim, discrete wavelet transform (DWT) and additive wavelet transform (AWT) in conjunction with the Mann-Kendall (MK) test are used and compared for the first time. Moreover, global wavelet spectrum (GWS) is employed to test the significance of the most effective periodic components. The sequential MK test is also used to determine the start or change points of trend in AIMF series. From the MK trend results, five stations showed a statistically significant (at a 5% level) negative trend for AIMF series and short-term periodic components (2 and 4 years) were found to be the most effective components, which are responsible for producing a real trend founded on the data series. The GWS analysis indicated that the most dominant components identified are significant. In addition, the MK-z values of the most effective periods derived from AWT generally showed a better agreement with MK-z value of original time series with higher correlation coefficient compared to those of DWT. The sequential MK graphs of the AWT-based time series also produced a better harmony with the sequential MK of the original data. Finally, the results showed AWT coupled with the MK provided a very efficient and accurate analysis for defining the most effective modes in the AIMF series and can be successfully used in any hydrological time series. (C) 2020 American Society of Civil Engineers.