Analysis of the 6 September 2015 Tornadic Storm around the Tokyo Metropolitan Area using a 3DVAR and Incremental Analysis Updates Coupling

In this study, we will present a preliminary analysis of the tornadic storm on 6 September 2015 around the Tokyo metropolitan area, Japan, using the Cloud Resolving Strom Simulator (CReSS) 3DVAR system coupled with an incremental analysis updates (IAU). In this storm, five wind damage were reported and at least two of them were identified as tornadoes. The Fujita scale of the strongest tornado was F1. This storm was observed by multiple X-band polarimetric Doppler radars and Doppler lidars, simultaneously. For this storm, the CReSS 3DVAR+IAU with Meso-Scale Model (MSM, NWP operated by the Japan Meteorological Agency) for background was used to analyze thermodynamical and dynamical fields of this tornadic storm.

On that day, a stationary front was laid on the Tokyo metropolitan area. Five wind damage occurred during about 2 hours (from 2020 LST, Local Standard Time: UTC + 9 hours, to 2210 LST), and three active convective cells spawned tornadoes and wind gusts. The analysis of the X-band polarimetric Doppler radar shows that there were persistent mesocyclones (vertical vorticity > 0.01 s^-1) in the three cells at 1 km AGL.

The 3DVAR+IAU analysis used radial velocities from multiple X-band polarimetric Doppler radars and Doppler lidars and radar reflectivity from multiple X-band polarimetric Doppler radars. The analysis grid spacing was 500 m for horizontal direction. The analysis period was from 1800 LST to 2400 LST, and 3DVAR+IAU was calculated every 10 min. The surface wind speed analyzed by 3DVAR+IAU (hereafter, V_3DVAR+IAU) was greatly improved than the control run (hereafter, V_CNTL). The root mean square error (RMSE) of V_CNTL was 3.2 m s^-1 (there were 87 observation stations in the analysis domain), whereas RMSE of V_3DVAR+IAU was 2.7 m s^-1.

The numerical simulation of this tornadic storm was calculated using above analysis for the initial condition. The horizontal grid spacing was 50 m and the lowest vertical grid spacing was 40 m. The simulation successfully reproduced several strong vortices near the surface. The maximum vertical vorticity of the strongest vortex was 0.36 s^-1. The analysis of the vorticity budget shows that the vertical vorticity was produced by the tilting term along the local boundary initially, and by the stretching term when the strong vortex appeared near the surface.