Declining runoff sensitivity to precipitation following permafrost degradation: Insights from event-scale runoff response in the Yellow River source region

Frozen ground, including permafrost and seasonally frozen ground (SFG), is a critical element of the cryosphere that strongly regulates hydrological processes in cold regions. It has been debated whether frozen ground degradation will make landscapes more, or less, sensitive to precipitation inputs; either outcome has profound implications for water resources and climate resilience. Using a data-driven approach based solely on observations, we quantify four decades of changes in event-scale runoff responses to daily precipitation in the source region of the Yellow River on the northeastern Tibetan Plateau. We apply Ensemble Rainfall–Runoff Analysis (ERRA), which infers hydrologic impulse responses directly from precipitation–runoff data without relying on model assumptions. This enables the assessment of nonlinear, nonstationary, and spatially heterogeneous hydrologic behavior across different frozen ground types and precipitation intensities. Results show that, relative to 1979–1998, the permafrost zone during 1999–2018 experienced a 47% reduction in peak runoff response per unit precipitation and a 32% decrease in the 25-day runoff coefficient, while the SFG region showed no substantial changes. The weakened runoff response in the permafrost zone, particularly under high-intensity precipitation (>10 mm d⁻¹), likely reflects enhanced infiltration and subsurface storage associated with thaw-induced deepening of the active layer. These findings highlight the power of data-driven approaches in detecting hydrological regime shifts and provide critical insights for drought mitigation and flood risk assessment in permafrost-affected regions.