The Structure of Descent in the Eye of Tropical Cyclones

We examine the structure of descent in the eye of numerically simulated tropical cyclones, and its relationship to the formation and maintenance of the warm core. In agreement with several previous studies, we find that mean descent tends to be much stronger at the edge of the eye than near the center. Schubert et al. (2007) found that this radial structure occurs theoretically for certain distributions of inertial stability, and further found observations of storms in which flight-level (~3 km) temperature was maximized at the edge of the eye. Preliminarily, we find that the concentration of descent at the edge of the eye occurs generally for most storms exceeding a relatively weak intensity threshold, and that this structure occurs throughout most of the troposphere. Nevertheless, the “warm ring” structure in temperature only occurs at low-levels and occasionally from 10-12 km, and we use a potential temperature budget to illustrate the reasons for this.

Near the center of the eye, the mean vertical velocity is often maximized near 12 km (~3-6 cm/s) during intensification, but the strongest warming is generally found to occur much lower (4-6 km). Several previous hypotheses on eye dynamics presume that descent is maximized at low-levels and decreases in magnitude with height; this is not consistent with our simulations. We also find that while the structure and magnitude of the mean eyewall updraft can be well-predicted by a linear vortex model forced by the mean diabatic heating, the structure and magnitude of descent in the eye can not. Reasons for this discrepancy are explored, as are its implications.

We use trajectories to show that under optimal conditions, it is possible for parcels to descend through much of the depth of the troposphere while remaining inside of the eye. We show that the instantaneous vertical velocity structure of the eye is dominated by high-amplitude transients (gravity waves) of up to several m/s, and it is therefore impossible to determine anything about the mean descent without averaging frequently sampled (< ~12 minute) data over long periods. Finally, the vertical structure of mean descent in the eye varies substantially even over longer time scales (12-24 h).