Integrated flood risk management based on urban policies remains challenging compared with infrastructure due to the unclear risk-reduction effects over time. To consider heterogeneity in social responses to urban planning measures, a previous study developed an agent-based household relocation model under flood risk (AHR-FLOOD) by combining flood-inundation and agent-based relocation models. However, accurate modelling of relocation incentives remains challenging. This study aimed to improve AHR-FLOOD by introducing spatial amenities based on transportation convenience and reflecting their impact on the decision-making processes of agents to analyze the development of different flood risk areas. AHR-FLOOD considering access to public transportation reproduced the spatial characteristics of the actual population and housing prices. The development of low-risk areas reduced flood exposure and resulting flood damage. However, this effect was less clear for low-income individuals, but the development of low-amenity areas had the potential to induce low-income population to move to safe areas. Chain migration was observed as a long-term effect of the spatial amenity policy. This study presents insights into the effect of transportation policies on flood safety for long-term spatial distance management in an agent-based approach with the rigorous modelling and validation of local amenity impact on household relocation choices.