Floods drive dynamic and deeply uncertain risks for people and infrastructures. Uncertainty characterization is a crucial step in improving (i) the predictive understanding of multi-sector dynamics and (ii) the design of risk management strategies. Current approaches to estimate flood hazards often sample only a relatively small subset of the known unknowns such as model parameters. This approach neglects the impacts of key uncertainties on hazards and dynamics, can drastically underestimate the tails of flood hazard probability distribution, and can result in poor decisions and outcomes. Here we assess whether and how a sequential Monte Carlo method can improve model hindcasts and projections. Specifically, we deploy a particle-based approach that takes advantage of the massive parallelization afforded by modern high-performance computing systems. We use this tool to analyze two specific questions: (i) What are the effects of accounting for more known unknowns on projected flood hazards? (ii) What are the key parametric uncertainties driving the uncertainties in projected flood hazards and risks?