Background <p>Current landslide hazard assessments often suffer from inadequate hazard characterization due to the neglect of parameter variability and reliance on a single dynamic indicator. To address this, this study proposes a probabilistic assessment framework that integrates parameter uncertainty analysis and dual dynamic indicators. Based on the Voellmy model, the framework simultaneously employs maximum momentum and maximum flow depth to characterize impact kinetic energy and burial effects, respectively. Furthermore, the uncertainty propagation of basal friction and turbulence parameters is quantified through Latin hypercube sampling, which is followed by the establishment of a Monte Carlo joint probability model and a three-level hazard zoning system.</p> Results <p>Application of this framework to the Sedongpu landslide demonstrates that the dual-indicator approach leverages complementary physical mechanisms to significantly optimize the identification of high-hazard areas. By providing a reliable multi-dimensional quantification of landslide hazards, this method addresses the deficiencies of traditional approaches, which are typically limited to binary reachability and univariate intensity metrics. Specifically, the integration of dual indicators yielded a performance improvement of nearly 10% in hazard identification compared to single-indicator methods.</p> Conclusions <p>proposed probabilistic framework successfully overcomes the limitations of single-indicator hazard assessments by integrating parameter uncertainty and dual dynamic indicators. This multi-dimensional approach provides a more accurate and reliable evaluation of landslide hazards, offering a solid scientific basis for differentiated disaster prevention and risk management strategies.</p>