Mitigating systemic risk in cascade hydropower generation: A high-dimensional vine copula framework for correlated inflow uncertainties

In the architecture of modern clean energy systems, cascade hydropower system function as the backbone of grid stability, playing a critical role in smoothing fluctuations from intermittent renewables and ensuring operational security. However, as global climate change intensifies, the uncertainty inherent in runoff forecasting is exhibiting increasingly complex nonlinear propagation and spatial superposition effects across multi-reservoir systems. These dynamics create systemic risks that are difficult to quantify using traditional linear models. We employed a C-vine copula model to precisely capture the “leptokurtic, heavy-tailed” characteristics and asymmetric spatial dependence structure of runoff forecast errors among hydrological stations within the basin. Subsequently, a power generation optimization scheduling model encompassing the lower Jinsha River cascade (Wudongde, Baihetan, Xiluodu, Xiangjiaba) and the Three Gorges Reservoir was constructed and solved via a Differential Evolution algorithm. Empirical results from daily-scale simulations reveal unique risk distribution patterns: (1) Spatial Dependency: The C-vine structure confirms that the Cuntan Station, serving as the basin's hydrological hub, exhibits significant upstream-tail (simultaneous flood occurrence) correlation risks with tributaries during high-water periods, indicating high spatial synchrony in extreme flood events; (2) Physical attenuation mechanism: Risk propagation between cascade reservoirs follows a “central aggregation, peripheral dissipation” pattern. The massive regulating reservoirs of Baihetan and Xiluodu absorb flow fluctuations from the upstream Wudongde, decoupling the upstream volatility from the downstream Three Gorges Reservoir, though they bear higher operational deviation risks themselves; (3) Resilience via Elasticity: Systemic risks are highly sensitive to the rigidty of scheduling constraints. Expanding the allowable deviation range from ±2% to ±8%, activates the joint regulation potential of cascade reservoirs, effectively mitigating over 90% of random disturbances. This study clarifies the distinct roles of reservoirs with varying regulation capabilities in risk management, providing theoretical support for transitioning from deterministic scheduling to a risk-oriented resilient scheduling paradigm.