Application of a Budyko-LMDI Framework to Evaluate Runoff and Sediment Attribution Across Multi-Hydrological Periods in the Western Loess Plateau

Abstract

The runoff and sediment processes in the Loess Plateau exhibit significant seasonal variability, yet traditional attribution methods often overlook water balance non-closure and precipitation phase transitions at seasonal scales. Utilizing hydro-meteorological data (1980–2019) from the Zuli River Basin, this study establishes an integrated methodological workflow by systematically integrating snowfall partitioning, the ABCD water balance model and the Logarithmic Mean Divisia Index (LMDI) method into the Budyko hypothesis. The ABCD model estimates seasonal water storage changes for physical consistency, while the LMDI method ensures a residue-free decomposition of runoff and sediment drivers. Results show an overall decline in both runoff and sediment across four hydrological periods. Accounting for hydro-meteorological input and storage estimation uncertainties via Monte Carlo analysis, runoff decrease in the hydrological year and wet season is predominantly driven by human-induced underlying surface changes, with contributions of 134.58% [95% CI: 107.82% to 163.17%] and 165.10% [95% CI: 145.16% to 186.44%], whereas dry-season runoff increase is primarily propelled by underlying surface regulation (179.59%) and enhanced snowfall (105.66%), while being substantially counterbalanced by seasonal water storage depletion (−163.88%). For sediment dynamics, the reduction in the wet season is predominantly driven by underlying surface changes (n, 99.44% [95% CI: 86.89% to 112.57%]), whereas during the hydrological year, dry season and snowmelt period, the decline is primarily dominated by decreasing sediment concentration (mC), with contributions of 61.30% [95% CI: 59.56% to 63.16%], 113.91% [95% CI: 110.41% to 117.89%] and 92.80% [95% CI: 90.71% to 95.08%], respectively. However, during the hydrological year, the role of underlying surface changes also remains substantial (54.07% [95% CI: 42.56% to 65.70%]). This study clarifies seasonally divergent driving mechanisms, underscoring the reshaping of hydrological processes by human activities and the crucial regulatory role of snowfall and storage dynamics during dry and snowmelt periods. The work offers a multi-scale scientific basis for tailored soil and water conservation on the Loess Plateau.