Assessing the driving process of meteorological and agricultural drought in the Yellow River Basin

The Yellow River Basin. This study examines six-month anomalies in precipitation (SPI), runoff (SRI), surface-layer soil moisture (SMI), and vegetation health (VHI), covering the period from 1990 to 2020. Multifractal detrended fluctuation analysis (MF-DFA) measures the time-dependent fractal memory of each index, and wavelet coherence shows the links between different drought types. MF-DFA shows that SMI is persistent at both seasonal and semi-annual windows. SRI goes from being persistent at 3 months to being anti-persistent at 12 months, and VHI shows a random-walk pattern at the seasonal scale before becoming persistent at longer scales. The average coherence between SRI and SMI is 0.55, with coefficients > 0.60 across a quarter of the basin and a significant-coherence area (PASC) > 50 % in 10 % of grids. Memory has the most direct effect on SMI, which has a mean coherence with its own past of more than 0.4 at all three timeframes. Our results collectively indicate that (i) fractal memory enhances soil moisture retention, (ii) hydrological drought serves as the principal pathway for meteorological deficits to impact the root zone, and (iii) vegetation stress exhibits a weaker and more heterogeneous response, particularly beyond agricultural areas. These insights enhance the comprehension of drought progression and establish a scale-sensitive framework for early warning throughout the Yellow River Basin. • Identified the fractal behavior of the drought processes at various timeframes. • Explored the scale-dependent control of fractal characteristics on drought. • The timeframes and spatial features influenced the fractal behavior of drought. • A more evident impact of fractal behavior occurs on the evolution of SMI. • The strongest coherence is found in the relationships between SRI and SMI.