An integrated framework for non-stationary hydrological drought assessment using time-varying parameter standardized streamflow index and time-varying threshold level method

Study region: Weihe River basin (WRB) in northern China. Study focus: In a changing environment, traditional drought assessment methods may not be applicable when assumptions of stationarity are violated. Accordingly, this study proposes a framework that incorporates the time-varying parameter standardized streamflow index (SSIvar) and threshold level method (TLvar) for the non-stationary hydrological drought assessment. Then, the SSIvar and TLvar methods are compared with time-invariant and transplantation parameter SSI (SSIinvar and SSItran) and TL (TLinvar and TLtran) to validate their effectiveness. New hydrological insights for the region: Validation results showed that SSIvarhas the highest Kendall correlation coefficients with standardized precipitation index (SPI) and soil moisture index (SSMI) at 0.81 and 0.78, respectively, outperforming SSIinvar(0.67and 0.62) and SSItran (0.68 and 0.63). The TLvar method behaves in the same way, indicating that the SSIvar and TLvar methods provide a more accurate assessment of non-stationary hydrological drought. Furthermore, the comparison results show that the mean duration and severity of hydrological drought in the WRB increased by 22.37 % and 13.72 % for SSIvar method and 34.69 % and 19.15 % for TLvar method from 1961-1990 to 1991-2020, respectively, revealing that hydrological drought in the WRB has aggravated over the past 30 years. Overall, the combined use of SSIvar and TLvar provides a comprehensive understanding of non-stationary drought, integrating qualitative (e.g., severity levels) and quantitative (e.g., streamflow deficits) measures.