[5] Model-related issues 2

[5-3] February 27, 17:00-17:20

Consideration of data assimilation for forecasting tornadoes using high-resolution ensemble forecasts

S. Yokota (Meteorological Research Institute), H. Seko (Meteorological Research Institute/JAMSTEC), M. Kunii (Meteorological Research Institute), H. Yamauchi (Observations Department, Japan Meteorological Agency), and H. Niino (Atmosphere and Ocean Research Institute, The University of Tokyo)


It is not possible to resolve tornadoes in operational numerical prediction models that have a horizontal resolution of several kilometers. If tornadogeneses depends strongly on the mesoscale environment, however, it may not be necessary to explicitly resolve tornadoes for its forecast. To clarify important ingredients in the initial state of the models for forecasting tornadoes, we performed ensemble-based analyses using 33-members high-resolution ensemble forecasts of the Tsukuba City F3 tornado on 6 May 2012. The horizontal resolution of the model was 50 m. The initial and boundary conditions were ensemble forecasts with 350-m horizontal resolution started from local ensemble transform Kalman filter analyses with 1875-m horizontal resolution, which assimilated four C-band radars and dense surface data. The results of backward trajectory analyses of parcels in forecasted near-surface tornado-like vortices show that the generation processes of the circulation of the vortices were not essential for determining the strength of the vortices. However, the mesoscale environment such as strength of mesocyclones at about 1-km height and near-surface humidity had strong correlations with vertical vorticity of the vortices. Thus, potential of tornadoes can be forecasted more reliably by improving numerical prediction of the mesoscale environment through a more sophisticated data assimilation.

  Presentation file: 05_3_S.Yokota.pdf