A framework for testing tropical cyclone hazard models

Abstract The field of tropical cyclone (TC) hazard modeling is in the midst of a rapid transition away from purely statistical techniques based on historical events to more physics-based approaches that can better account for the climate change that has already occurred and that will continue into the future. As such models proliferate, it is important to test them against TC observations and compare them to each other. Here a framework is proposed for comparing hazard model predictions to metrics derived strictly from historical (‘best-track’) observations, though in the future metrics more directly related to damage should be incorporated. While a specific set of metrics and statistical tests is presented here for illustrative purposes, it is hoped that the TC hazard community along with industry and government interests will convene to agree on a set that will serve as a benchmark for testing and intercomparing TC hazard models. This should allow for more rapid refinement and improvement of such models.