Abstract: Flood forecasting in ungauged reservoirs downstream of inflow control stations is hindered by scale effects and numerical inaccuracies arising from large disparities between main stem and tributary drainage areas. To address these issues, a spatial multi-scale nested simulation method based on the TOpographic Kinematic Approximation and Saturation-Infiltration Double Excess （TOKASIDE ） model was developed and verified in the Yuecheng Reservoir Basin. Scale effects in gauged basin were analyzed and cross-scale parameter transformation relationships derived from fractal and geomorphic theories were established, forming a spatial multiscale-nested TOKASIDE framework. The results show that: ① in terms of scale effects, simulated discharge varies systematically with resolution, decreasing in the 100—500 m range and increasing in the 600—2000 m range; ② applying scale transformations to the surface and channel roughness reduce systematic bias across resolutions; and ③ the application to the Yuecheng Reservoir basin indicates that, for two historical flood events, the Nash–Sutcliffe efficiency （E_NS） values of the multiscale scheme are both greater than 0.7, and the peak discharge error is reduced by approximately 27% compared with single-resolution （1000 m） simulation. The proposed method resolves parameterization and scale effects in ungauged reservoir reaches, significantly improving accuracy and reliability, while providing technical support for reservoir flood-control operations.