Code and Dataset for "Journal of Advances in Modeling Earth Systems Understanding tropical cyclone frequency biases in a global storm resolving model"

The code and X-SHiELD's TC seed dataset for "Understanding tropical cyclone frequency biases in a global storm resolving model" Abtract: Global storm resolving models (GSRMs) provide an excellent testbed for understanding the role of small-scale processes in tropical cyclogenesis. In this study, we assess tropical cyclone (TC) genesis and frequency in the Geophysical Fluid Dynamics Laboratory's eXperimental System for High‐resolution prediction on Earth‐to‐Local Domains version 2021 (X-SHiELD v2021). Compared to observations, X-SHiELD produces significantly fewer TCs, albeit with realistic representations of both the number of pre-TC disturbances (TC seeds) and the mean state relative to ERA5. We find that the TC seed-to-TC transition rate is underestimated in X-SHiELD because the spin-up of TC seed vortex is inefficient. The inefficient TC seed spin-up in X-SHiELD is associated with a weaker-than-observed low-level inflow, which is coupled to mesoscale ascent near the center that is too weak and too top-heavy. The biases in vertical velocity are associated with a dry bias at low levels, which is caused by the overly active shallow convection scheme. Our results underscore the importance of subgrid scale processes, particularly low level mixing through shallow convection, in the representation of TC genesis process even in kilometer scale models.