Development of Building Flood Fragility Curves in Bali Through Bayesian Inference and Two-Dimensional Hydraulic Simulation

Abstract

The riverine areas of Bali face complex challenges from increasing climate-change impacts, population growth, rapid land-use change, and rising economic activity. These conditions have exacerbated flood risk, particularly along the Tukad Badung River. Effective flood mitigation requires appropriate risk assessment, particularly accurate quantification of hazard and fragility functions. Given limited damage data and uncertainty in observed inundation depth, a locally calibrated flood fragility curve should account for uncertainties in inundation depth from both observed data and numerical models. This study develops empirical flood fragility curves for buildings in the Bali region using a Bayesian approach. Building damage data and associated depths were obtained from field surveys during the September 2025 flood event, while hydrological and hydraulic simulations evaluated inundation depth at the surveyed points. Using both numerical and observed depths, fragility curves were developed through Bayesian inference, explicitly representing uncertainty in the observed damage state and modeled hydraulic parameters. The results show that combining hydraulic simulation with Bayesian statistical analysis provides a more reliable representation of flood vulnerability than conventional approaches. The proposed method can support more adequate flood-risk evaluation and disaster countermeasures in Bali.