Managing Cascading Flood Risks in Reservoir Networks: A Copula-Bayesian Framework for Ensemble Forecast Error Propagation and Risk Equity

Flood ensemble forecasts introduce spatiotemporally heterogeneous risks to multireservoir systems. This study proposes a novel framework integrating: (1) a Copula–Monte Carlo simulation model generating synthetic ensemble forecasts while preserving multivariate error correlations; (2) a risk parity scheduling model balancing reservoir-specific flood risks; and (3) a Bayesian network-based risk evolution analysis quantifying spatial risk propagation. Applied to the Huaihe River's Wangjiaba system (8 reservoirs, 4 flood sections), results demonstrate:1)The simulation model preserves temporal (correlation >0.75) and inter-member error correlations but weakens spatial correlations. 2) Bayesian analysis identifies Suyahu and Shishankou reservoirs as most vulnerable (probability ≈30%), primarily due to high forecasted inflows and uncertainties. 3)The framework provides a holistic solution for managing ensemble-forecast-induced risks in complex reservoir networks.