Abstract: To address the insufficient representation of monthly inflow uncertainty and the difficulty in identifying dry-scenario water shortage risk and wet-scenario regulation pressure in annual inter-basin water transfer planning, this study proposes a framework integrating monthly runoff probabilistic forecasting, inflow scenario generation, and water-balance checking. Observed monthly runoff is used as the unified evaluation target. Monthly mean runoff estimated from the training period is adopted as a seasonal baseline, and probabilistic residual modelling is then conducted. A bidirectional gated recurrent network with an attention mechanism and non-crossing quantile constraints is developed to generate dry, normal, and wet inflow scenarios. A case study of Sanhekou Reservoir shows that the Nash-Sutcliffe efficiency coefficient in the test period is 0.3608, and the coverage probability of the 90% prediction interval is 0.9275. With an illustrative annual transfer target of 500 million m3, the dry scenario provides 181.18 million m3 transferable water and a shortage of 318.82 million m3. The normal and wet scenarios both meet the target, while the spill or forced-release volumes are 114.14 million m3 and 1.42659 billion m3, respectively. The results indicate that the proposed framework can provide inflow boundaries and risk indicators for annual water transfer planning and rolling adjustment.