A numerical simulation of early phase of tropical cyclogenesis associated with multiple tropospheric vortices

In the early stage of the tropical cyclogenesis (TCG) convective clouds organize cloud cluster, in which strong TC initial vortex is sometimes formed and sometimes not. The processes involved in the very early stage of TC vortex formation are not yet well understood. Bister and Emanuel (1996) suggested that potential vorticity of preexisting MCV in mid-troposphere develops downward to the lower troposphere, organizes convergence at PBL, and let convection activate. Ritchie and Holland (1997) investigated the formation of typhoon Irving and also suggested that the PV at mid-level seems to enhance to low-level. In contrast to these ‘top-down' development cases, Kieu and Zhang (2008) investigated the organization of the tropical storm Eugene and they showed that the vortex seems to develop from the lower to the upper level after the merger of two MCVs. So far, it is not obvious which mechanism ‘top-down' or ‘bottom-up, or the both are fundamental to the initial organization of TC vortex. Since the meso-scale processes over the tropical ocean is difficult to be observed, the alternative approach is to analyze the numerically simulated TCG. We run WRF-ARW 3.0.1.1 to reproduce the Typhoon Francisco in 2001 and investigated the initial formation of the TC vortex closely. The Francisco was first recorded in JTWC best track data as ‘TD' on at 13.5N 166.5E on 0600 UTC 17 Sep. 2001. It developed to ‘TS' intensity at 14.6N 161.5E at 1200 UTC 19 Sep. 2001. In the numerical simulation quadruple nested computational domains is employed in which the smallest grid resolution is 1 km (448×439 grids). The model is initialized at 0000 UTC 15 Sep. 2001, 54 hours before the first identification as TD, using NCEP/FNL data. The data is also used as lateral boundary condition of the largest domain. The model was integrated for 6 days and a TD was successfully generated about the same timing as Francisco, but the location was about three degrees to the north. Since the lateral boundary where the NCEP/FNL is applied is far from the location of the TCG, we assume that the TCG was not affected by the boundary condition and analyze the detailed meso-scale processes involved in the simulation. A weak MCV, which diameter is approximately 400 km and is characterized by potential vorticity less than 1 PVU, exists in mid-troposphere near melting level (6 km). This weak MCV developed into the stronger one with potential vorticity greater than 2.5 PVU, and the diameter shrank to 100 km approximately till 00UTC, 16 Sep., 30 hours before the TD genesis. Weak downward motion existed beneath the MCV. On the other hand, several cyclonic vortices were generated in lower troposphere after the start of integration. The horizontal scale of these low-level vortices (LLVs) is from 40 to 50 km in diameter. Above these LLVs convective updraft existed, which form slender tower of strong potential vorticity region. The mid-level MCV and LLVs exist in different horizontal locations initially; eventually they approach each other, presumably due to the vertical shear of horizontal wind. When LLVs migrated under the mid-level MCV, LLVs merge each other and was strengthened. The merger of LLVs continued and a monolithic PV area with many convective clouds was formed at 00UTC, 17 Sep. The horizontal wind speed with this vortex reached to 13 m/s at 06UTC. Firstly, we see the vertical development of MCV. The figure (a), shows the time-height plot of potential vorticity for MCV from 0000 UTC 16 to 0000 UTC 17. The potential vorticity is computed as an average over the disk area of 100 km diameter around the center of MCV. Figure (b) and (c) are the same but for two prominent LLVs. For LLVs the averaging dimension was set as 45 km in diameter around the center. In these figures, contour shows potential vorticity, and color shade does relative vorticity. From (a), it is seen that strong potential vorticity area of MCV expands from mid- to low-level, especially at 1400 UTC 16. For the LLV1, strong potential vorticity locates below 3 km height initially, and it expands upward after 12UTC of Sep. 16. Same tendency is seen in relative vorticity. For LLV2 the upward expansion of strong potential vorticity area is seen in the same period. From this analysis of numerically simulated TVG of Typhoon Francisco, it is inferred that both ‘tip-down' and ‘bottom-up' development takes place but in different horizontal scale. The ‘top-down' development is seen in MCV scale and the ‘bottom-up' development is seen in the smaller scale.