Structure and formation of convection of secondary rainbands in a simulated typhoon Jangmi (2008)

Abstract

Secondary rainbands in tropical cyclone are relatively transient compared with the quasi-stationary principle rainbands. To have a better understanding on their convective structure, a cloud-resolving scale numerical simulation of the super typhoon Jangmi (2008) was performed. The results suggest that the convections in secondary rainbands have some distinctive features that may not be seen in other types of rainbands in tropical cyclone. First, they have a front-like structure and are triggered to form above the boundary layer by the convergence of the above-boundary outflow from the inner side (warmer) and the descending inflow (colder) from the outer side. These elevated convections can be further confirmed by the three-dimensional backward trajectory calculations. Second, due to the release in baroclinic energy, the lower portion of the mid-level inflow from outside may penetrate into the bottom of the convection tower and may help accelerate the boundary layer inflow in the inner side. Third, the local maximum tangential wind is concentrated in the updraft region, with a lower portion which is dipping inward. Tangential wind budget analysis also suggests that the maxima are mainly contributed by the updraft advection, and can be advected cyclonically downstream by the tangential advection.

Appendix

1.1 Semi-automatic rainband tracking algorithm

In this study, the rainband axes were determined by the local reflectivity maximum in each radial cross-section. When the initial and end points of a rainband are given, the locations of the local reflectivity maxima can be successively determined by iterative calculations, with the first-guess positions equal to the results of the previous calculations. The detailed steps are as follows:

• 1. The reflectivity field at 2-km height is interpolated into a polar coordinate. The origin point is at the vertically weighted average of the vorticity centroid. The resolutions are 1 km and 0.5° in radial and azimuthal directions, respectively.

• 2. The initial and end positions of the rainband and the radial searching range are then determined artificially. In this study, the range was from 3–5 km inside to 5–10 km outside the rainband axis. If there is a gap between the convective cells or if two rainbands are too close to each other, extra anchor points are required.

• 3. Based on the first-guess position and the searching radius, the iterative calculations are repeated in each radial cross-section until the results convergence. This result is identified as the rainband axis in this radial cross-section, and will be considered as the first-guess position for the next radial cross-section (in cyclonic direction) calculation. This process continues until all the reflectivity maxima of the rainband have been determined. If there is an given anchor point, then the calculations will jump to the next radial cross-section.

• 4. To decrease the dithering in the rainband axis within the gaps (reflectivity maxima ≤ 40 dBZ) between convective cells, the axes lines were replaced by linear extrapolation from both ends.