Tropical cyclones (TCs) and their associated hazards have been notoriously difficult to resolve in climate models owing to a lack of horizontal resolution. While recent model advances have pushed grid spacings to ~25km, which permit TCs, simulated high-end winds and precipitation continue to be biased low. Information at the regional stakeholder level has therefore relied on statistical downscaling techniques based on model projections to ascertain changes in risk.

This talk discusses the potential merits of an event-level dynamical downscaling strategy: short, high-resolution "storylines" to evaluate and communicate climate risk. We leverage lower-resolution, less expensive, global Earth system model simulations to generate gray swan tropical cyclones (physically plausible but historically unrealized events extracted from multi-decadal climate ensembles). We then reinitialize these individual storms in the same model at 3km grid spacing, demonstrating increased structural fidelity of events when compared to observations. Validation of historical storms against 4km gridded NEXRAD data indicates that this km-scale grid spacing more accurately produces precipitation exposure at timescales shorter than six hours. We discuss scientific insights gained from these simulations, the strengths and limitations of this approach, and the potential benefits of using them to communicate with stakeholders and the general public about "unforeseen" hazards.