Tropical cyclone sensitivity to initialization in semi-idealized simulations

Previous work has demonstrated that cloud microphysics and related model physics can have a material effect on tropical cyclone (TC) motion in both real-data and "semi-idealized" simulations (e.g., Fovell and Su 2007; Fovell and Boucher 2009). The semi-idealized experiments were based on the real-data WRF model (the "real-ideal hybrid") and high-resolution simulations have revealed that assumptions relating to size distributions, hydrometeor speciation, and interaction with radiation strongly influence storm symmetric (Fovell et al. 2009) and asymmetric (Fovell et al. 2010) structure, modulating TC speed and direction. These experiments have utilized TCs created organically within the model, through the introduction of a synoptic-scale buoyancy perturbation in a calm, otherwise horizontally homogeneous atmosphere.

The previous work varied model physics for a single initialization. The present study (Cao et al. 2011) employs the real-ideal hybrid model to examine the role of the initialization on TC motion and structure, making use of brute-force vortex bogussing. Vortices of various outer wind shapes, determined by the modified Rankine profile decay parameter (alpha), are inserted into the otherwise horizontally homogeneous initial environment. Smaller values of alpha imply relatively larger winds beyond the core. The radius of maximum wind (RMW) is also varied.

Unsurprisingly, TCs provided with stronger winds at larger radii acquire faster motions, quickly resulting in a tremendously large track spread. The model TCs, however, do not retain the initially supplied wind profiles very long. Instead, they evolve quickly towards a common shape that is determined by the model physics. This homogenization casts doubt on the ability of models to hold onto and propagate forward information supplied,at the initialization by advanced data assimilation techniques or parameterized vortex wind profiles. The bubble and bogussed TCs also developed markedly different heating patterns, which appear to help explain why the artificially-established storms tended to move, on average, about three times faster.