AGCT: a hybrid model for identifying abrupt and gradual change in hydrological time series

Abstract

A new hybrid model called the AGCT model has been proposed for identifying an abrupt and gradual change in a hydrological time series. This hybrid model combining the over-whitening procedure, Pettitt test, and Mann–Kendall test can measure abrupt and gradual changes in hydrological series with mixed-mode shift and weight the attributions of two types of changes to the shift. In addition, four types of shifts were defined as gradual change, gradual-leading change, abrupt-leading change, and abrupt change in this study to describe the shifts observed well, which can provide more detailed and useful insights into changes in hydrological time series. Using the Wuding River Basin in China as a case study, the results indicated that the AGCT model can effectively separate the abrupt-change component from a complex shift, measuring the abrupt and gradual changes embedded in a streamflow time series, and further determine the shift type of the series. In addition, it is considered as a feasible method to steady the nonstationary streamflow data for frequency analysis.