<p>In the context of climate change, flooding presents a significant challenge for urban areas. The situation is worse with the increasing density of impervious surfaces and population. As an essential part of flood risk management, flood risk assessments can provide vital information to help local decision-makers and related departments formulate effective strategies. The indicator-based approach, an established method in hazard risk assessments, is widely used as it can allow for a systematic analysis of risk components. Although this approach is helpful, it is limited in handling new data and uncertainty in dynamic urban environments. To address these limitations, we developed a Bayesian network (BN) model based on an established indicator-based framework. This model allows for the dynamic incorporation of new evidence and provides explicit probabilistic quantification of uncertainties in the risk assessment process. This novel method combines the systematic analysis of the indicator-based approach with advanced computational capabilities to assess flood risks more dynamically. Data from eight historical flood events in Houston, USA, between 2000 and 2019 were used to train and test the BN model. By comparing the indicator-based approach and the BN model, 83.71% of the results were found to be identical. The flood assessment shows that during Hurricane Harvey, the western and northeastern regions of Houston faced the highest flood risks, while the central areas experienced relatively lower flood risks. In addition, the sensitivity analysis shows that three nodes under the exposure part, the exposed population, exposed forest, and exposed water area, are the most impactful factors.</p> Graphical abstract <p></p>