Conditional Design Flood Hydrograph Derivation Considering Spatially Heterogeneous Rainfall and Nonlinear Nonstationary Routing

Abstract. Design flood hydrographs are jointly influenced by the spatial distribution of rainfall, antecedent wetness states, and catchment runoff generation and routing responses. Taking the XDY Basin as the study area, this study develops a conditional design flood hydrograph derivation framework that accounts for spatially heterogeneous rainfall and nonlinear nonstationary routing. Ensemble Rainfall-Runoff Analysis (ERRA) is used to identify rainfall-runoff responses under different antecedent wetness states, net rainfall intensities, and spatial subregions, based on which state-dependent unit hydrographs are constructed. Joint frequency analysis of 96 h total rainfall depth (P96) and maximum 1 h rainfall intensity, together with a common representative rainfall pattern and spatial-response correction, is then used to derive design flood hydrographs under lower (L) and higher (H) antecedent wetness states. For a given return period, the L and H scenarios use the same design rainfall input. The results show that higher antecedent wetness and greater net rainfall intensity produce stronger runoff responses and earlier flood peaks. In the incremental model-structure comparison, the mean Nash-Sutcliffe efficiency (NSE) increased from 0.744 to 0.874. Compared with the L scenario, the H scenario increased peak discharge by 54.4 %–100.0 %, advanced the time to peak by 19–27 h, and increased flood volume by 26.6 %–59.9 %. Event-level paired Bootstrap analysis indicates that statistical uncertainty is concentrated mainly around the rising limb and main flood peak, while the effects of antecedent wetness exhibit good sample stability. The proposed framework provides process-based information for design flood review, flood-control safety assessment, and flood risk analysis.