Physically-based path clustering of ensemble forecasts for extratropically transitioning tropical cyclones

Ensemble forecasts are designed to span the range of future atmospheric states given the known uncertainties in an initial state. Summary forecasts of tropical cyclone [TC] characteristics are derived from these ensemble forecasts; public forecasts issued and verified include geographic track, intensity and wind radii; information on the evolution of the 3D structure of the TC is critical for systems undergoing extratropical transition [ET].

We characterize forecasts of the lifecycle evolution of the ET in terms of both its track (, ) and its structure; the cyclone phase space (B, VTL, VTU) provides a compact summary of the structure. We demonstrate that path clustering yields physically meaningful partitions of ensemble forecasts throughout the ET when applied both frameworks (track and CPS). We contrast the two resulting partitions to explore the treatment of ambiguous forecasts.