Automated tornadic vortex wind retrievals from radar observations

A variational method is developed and automated with data quality-control procedures to retrieve tornadic vortex winds from radar radial-velocity observations on each tilt of radar scan. This method has the following features: (i) The retrieval domain is centered at and moving with the detected vortex center on each tilt of radar scan. (ii) Vortex-flow-dependent background error covariance functions are formulated for the streamfunction and velocity potential in the moving frame. (iii) The square root of the background covariance matrix is derived, using the convolution theorem, from each covariance function and is used to construct the solution in a concise form in the cost-function. The method has been successfully applied to tornadic mesocyclone observed by the operational KTLX and KFDR radars on 24 May 2011 and to the EF5 tornadic vortex observed by the KTLX radar that struck Moore in Oklahoma on 20 May 2013. The method is also being extended to analyze three-dimensional tornadic vortex winds by using radial-velocity observations on all different tilts from either single radar or multiple radars with the mass continuity equation satisfied exactly in a strong-constraint form. The results will be represented at the conference.