With a minutely sampled proximity sounding at 1400 LST on 21 July 2012 at Beijing station (its surface level was replaced by a nearby surface observation), which was just ~ 20 km to the west of and ~ 20 minutes before tornadogenesis, the model faithfully captured the observed features from the motion and morphology of the supercell and its mesocyclone to the appearance and shrinking of descending reflectivity core (DRC), and more importantly, to the genesis, maintenance and demise of the tornado. The simulated tornado well captured damage-survey-deduced tornado track, lifespan and intensity variation, as well as its relationship with DRC.
Results showed that the tornado was generated through upward development of a near-surface vertical vorticity column likely through the stretching by updraft associated with the mesocyclone, and downward development of negative pressure perturbation collocated well with the stretched vertical vorticity column, both at tornadic scale (Fig. 1). Then the tornado experienced a replacement of the near-surface vorticity column and the associated negative pressure perturbation, well corresponding to the observed weakening and re-intensification of tornado damage and the transition from a linear to sinusoidal tornado track. The replacement of the near-surface vorticity column happened when the vorticity column became dislocated from the mesocyclonic updraft center while a new vorticity column formed nearby, rapidly got stronger and wider and coupled with the mesocyclonic updraft, resulting in the strongest ground-relative wind at the surface during the whole process, consistent with the observed most severe damage that caused fatality as the tornado re-intensified.
The tornado decayed right after the dislocation of the second vorticity column from the updraft center of the weakening mesocyclone. Though there was a third vorticity column developed underneath the mesocyclone, the stretching was strongly inhibited by the occlusion downdraft near the tornado at low level, the connection between the near-surface vorticity column and the mid-level updraft was cut off, thus the whole system demised eventually.
These replacements of near-surface vorticity column in the lower-part of the whole simulated vorticity column may also provide an alternative explanation for the first-linear-then-sinusoidal tornado track other than the traditional regarding that a tornado rotates around the mesocyclone center as one single vorticity column.
Examinations on the generation of near-surface rotation, the relationship between the tornado evolution and the DRC, the thermodynamic property of the tornado inflow and the occlusion downdraft are also underway.