Consistent with previous studies, in the presence of shear, RI onset tends to occur immediately after the tilt between the low- and mid-level centers becomes small and the vortex is nearly aligned. In our simulations, alignment occurs systematically earlier for broader vortices, and hence RI onset is also earlier. But it is unclear why alignment occurs earlier. One reasonable hypothesis could be that broader vortices initially tilt by a smaller amount in response to shear, but we show that this is not actually the case, as the initial tilt actually is very similar for different vortex profiles in the same shear. We have found that the distribution of convection (which occurs partly in response to the shear) begins to differ early in the simulations (long before alignment and RI), with a greater areal coverage of convection the broader the vortex. We hypothesize that the persistence of convection over a large area allows the broader vortices to slowly intensify in shear, which in turn hastens the alignment and RI onset times. We are currently investigating what specifically causes the broader vortices to have a larger and more persistent region of strong convection. One hypothesis is that the vortex response to the same tilt differs among the simulations, and through a set of analogous dry simulations, we show that the spatial distribution of tilt-induced ascent/subsidence is indeed broader for the broader vortices, and this may be a mechanism by which a feedback occurs in the full-physics simulations. However, there are other potential contributing factors that we are also examining, including the frictionally-driven ascent (which also varies with the vortex profile) and differences in the spatial coverage of surface enthalpy fluxes. In this work, we attempt to separate these different influences, in order to determine the ultimate cause for the sensitivity of RI onset timing to vortex structure.

