Polarimetric Doppler Spectra in Southern Plains Tornadoes Computed from Mobile Radar Data

Abstract

Doppler wind velocity estimates depend on the assumption that scatterers in the atmosphere are moving at the same velocity as their surrounding air molecules, as the radio waves used in weather radars do not significantly backscatter off air molecules. Most of the time, this assumption is valid, but in tornadoes, wind speeds become high and the airflow exhibits substantial curvature. Larger and more massive scatterers accelerate more slowly under the force of the wind and are also subject to an increased centrifugal force. Thus, these scatterers do not move at the same velocity as the wind, and velocity estimates can be subject to large errors/biases. In order to study tornado dynamics, errors in wind fields must be diagnosed and minimized. This can be accomplished by spectral analysis of raw, dual-polarization, in-phase and quadrature (I/Q) data. Since 2018, efforts have been made using the University of Oklahoma's rapid-scan, X-band, polarimetric, mobile Doppler radar to collect I/Q data in tornadoes and verify these velocity errors. Through Fourier analysis of I/Q signals, dual-polarization spectral densities (DPSDs) can be computed, which leverage the distinct polarimetric characteristics of rain and debris to separate their respective motions. In 2023 and 2024, RaXPol collected I/Q data from within three tornadic vortices, with the last one, on May 23, 2024, being the most comprehensive. This dataset was collected in an EF-2 tornado using a scanning strategy involving multiple low-level scans through the debris cloud of the tornado. The radar employed a slow azimuthal rotation (6 degrees per second) with sector scanning, achieving over 650 pulses per degree of azimuth. The spectra were evaluated for quality using a convergence metric that optimizes parameters for DPSD computation and assesses the requirement for slow scanning. With computed DPSDs and an automated fuzzy logic debris classification algorithm (DCA), spectra were analyzed, and areas of strong velocity bias were identified, consistent with numerical simulations. These results demonstrate the potential for improved tornado dynamics retrieval through spectral polarimetric analysis.