Accurate predictions of wave-driven flooding are essential to manage risk on low-lying, reef-lined coasts. Models to provide this information are, however, computationally expensive. We present and validate a modeling system that simulates flood drivers on diverse and complex reef-lined coasts as competently as a full-physics model but at a fraction of the computational cost to run. This development paves the way for application in large-scale early-warning systems and flood risk assessments.