Integrating vulnerability processes into a sociohydrological model

Introduction Sociohydrological models have advanced understanding of human -flood feedbacks and the non-stationarity of flood risk, yet vulnerability is often treated implicitly, homogeneously, or as a static parameter. Empirical evidence shows that vulnerability is multidimensional, socially differentiated, and evolves independently of short-term hazard experience, a dynamic that requires further exploration. Methods This study develops an enhanced human-flood feedback model in which flooding, exposure, sensitivity, adaptive capacity, vulnerability, and flood losses co-evolve over time. Vulnerability is explicitly represented as a dynamic internal state variable, consistent with the IPCC Sixth Assessment Report. Model experiments under identical hydrological forcing were conducted and complemented by a Sobol sensitivity analysis. Results and discussion Explicitly representing vulnerability fundamentally alters long-term risk trajectories. While short-term flood responses remain synchronous, the enhanced model exhibits higher exposure but lower flood losses, reflecting a decoupling between fast flood-driven dynamics and slower social processes. Flood losses emerge as a nonlinear function of hazard, exposure, and vulnerability, with the persistence of adaptive actions exerting the strongest control on long-term outcomes. These results highlight vulnerability as a slowly evolving social condition shaped by institutional learning and adaptation fatigue. Explicitly incorporating vulnerability into sociohydrological models provides a more realistic basis for long-term flood risk assessment and governance under increasing exposure and climatic uncertainty.