An integrated Bayesian hydrodynamic framework for quantifying rating curve uncertainty in design flood estimation and...

Uncertainty in river discharge, particularly stemming from rating curve limitations, remains underexplored despite its direct influence on hydrologic and hydrodynamic modelling outcomes. The study offers a framework that integrates a stationary Bayesian rating curve model (BaRatin), L-moment-based flood frequency analysis (FFA), and a 1D-2D (one-dimensional and two-dimensional) coupled hydrodynamic model for enhanced design flood estimation and flood hazard mapping in Ghatal town, India. Based on discharge measurements and prior hydraulic knowledge, the BaRatin is developed for the Shilabati River. Discharge estimates from the most probable rating curve, along with credible intervals (2.5 % and 97.5 %), are propagated through FFA, hydrodynamic simulations, and hazard mapping to account for uncertainty. FFA in the Shilabati River reveals that 8-year floods inundate the left bank, 17-22-year floods can breach the right bank, and 100- and 200-year floods can overtop the embankments. HEC-RAS model simulations of past floods accurately replicate observed water levels and inundation extents from embankment failures, resulting in high calibration and validation accuracy. The 100-year flood simulations indicate that the Shilabati and Rupnarayan rivers are likely to overflow their banks, resulting in widespread inundation in almost all parts of Ghatal (51.90 km²), with an estimated range between 52.217 km² and 48.844 km² under the 95 % credible interval. The study recommends raising the Shilbati's right bank embankment by 0.45 ± 0.3 m and the left bank by 1.95 ± 0.3 m to mitigate the 100-year flood. This framework provides a robust methodology for enabling better-informed infrastructure design and floodplain management under uncertainty for data-scarce regions.