A realistic flood risk assessment is important for rivers where the flood protection infrastructures are dictated by varying return periods. For rivers in Japan, design return periods for flood protection infrastructures range up to 200 years. Large-scale flood hazard mapping increasingly relies on global river models, but these models often lack explicit representation of flood protection levees. In this study, we extend the Global River Model (CaMa-Flood) by integrating levee parameters and applying frequency analysis to simulated flood volumes (the cumulative amount of water exceeding channel storage) and downscaling them to high resolution while explicitly accounting for topographic variability and levee protection.</p> <p>Levees are represented through heights and fractions, with fractions derived from distance to the river centreline and heights refined by simulations. The method applies both to current simulations, using modelled flood volumes directly, and to future hazard assessment, where frequency analysis of annual maxima provides return-period volumes. These volumes are redistributed to high-resolution unit catchments using terrain data and physically constrained by storage availability.</p> <p>The results show that integrating levee protection reduces simulated flood volumes, with 10–15 % reductions across most return periods in grids containing levees. This reduction reflects the confinement of floodwaters within levee-protected channels, which limits floodplain storage and lowers overbank volumes. At the unit catchment scale, flood extents are also reduced depending on levee fraction and topography. Levees effectively confined floodwaters during moderate to high events, while their influence diminished at extremes where overtopping or volume overestimation became prominent. Findings demonstrate that the levee-integrated downscaling approach captures spatial variability in protection effectiveness, offering a more realistic representation of flood hazard across diverse conditions. By combining hydraulic modelling, frequency analysis, and levee integration, this study provides a comprehensive framework for flood depth mapping, supporting improved resilience planning and basin management.