A non-stationary impactquant framework for assessing the human activity impacts on hydrological drought in the Upper Red River

Hydrological drought (HD) in dynamic environments, particularly complicated by climate change and human activity, challenges traditional stationary-based research approaches. These challenges are particularly pronounced in the Upper Red River Basin characterized by complex topography and extensive human activity. To these issues, this study introduces non-stationarity in determining drought propagation time and develops a Non-Stationary ImpactQuant Framework (NSIQF) to assess human activity impacts on HD's frequency and propagation processes. Implementing non-stationarity involves using the Generalized Additive Models for Location, Scale, and Shape to construct the Non-stationary Standardized Runoff Index (NSRI), which is employed for attribution analysis and determining drought propagation time. We systematically quantified human activity impacts by comparing NSRI with drought indices and propagation time derived from reconstructed natural runoff. Our findings indicate that the NSRI is more adept at capturing drought characteristics in changing environments than the traditional Standardized Runoff Index. The human activity impacts on HD in the study area exhibits distinct temporal characteristics. Human activity contributed to 42.64% of the HD at 12-month scales, and 29.28%, 17.68%,19.14%, and 24.82% in spring, summer, autumn, and winter, respectively. They prolonged the propagation time from meteorological drought to HD by 1–5 months. Land use change is a potential primary anthropogenic factor. Effective management and strategic interventions for predominantly agricultural activity constitute viable solutions to alleviate regional drought and enhance drought resistance.