The Importance of Bringing Tropical Cyclone Science to Bear on Long-Term Hazard Risk

While numerical forecasts of tropical cyclone have improved rapidly in recent decades, far less attention has been focused on bringing numerical models to bear on long-term tropical cyclone risk. As a result, most extant TC risk assessments are based on historical records. In this talk, I will demonstrate that such records are far too short for making robust long-term projections, and that climate change has already rendered such records largely irrelevant for assessing current, let alone future, risk. Poor modeling of current and future risk may have social consequences as severe as those arising from poor real-time forecasts. Global climate models are still far too coarse to simulate TCs with any fidelity, and while mesoscale global modeling is now feasible, such models cannot be run long enough into the future (or with enough ensemble members) to provide reliable long-term risk estimates. I will show that the advent of advanced downscaling techniques, including artificial intelligence, are beginning to change this landscape and discuss what might be done to accelerate interest in this problem among TC modelers.